Foundation Model

Wireless foundation models offer a path toward reusable channel state information (CSI) intelligence for sixth-generation (6G) systems. However, existing generic-backbone adaptation and CSI pretraining methods often treat CSI as task tensors rather than propagation-conditioned channel responses, thereby failing to capture the intrinsic time-frequency-spatial geometry of wireless environments. This paper presents a channel-adaptive roadmap toward CSI-native foundation models, proposing a unified framework that aligns pretraining, positional modeling, and attention control with three channel requirements: scale-aware heterogeneous exposure, physical time-frequency-antenna coordinates, and correlation-bounded token interaction. Extensive experiments demonstrate the superiority of the proposed framework across three dimensions: zero-shot generalization, reducing NMSE by more than 4 dB across spatial-temporal-frequency tasks; scale extrapolation, yielding up to a 5.4 dB gain under 8 times unseen antenna scaling; and inference efficiency, accelerating mobility-aware processing by up to 18.8%. A system-level evaluation with Sionna SYS further shows that the proposed framework uses only 7.01% of dense-pilot overhead, reaches -18.64 dB average NMSE, and improves average net spectral efficiency by 36.6% over dense LMMSE and 15.5% over WiFo, indicating that CSI-native representation learning can support pilot-efficient radio access.

Self-supervised foundation models have shown strong promise in medical imaging. However, existing MRI foundation-model studies have primarily emphasized segmentation and dense prediction tasks, while systematic investigation of self-supervised foundation models for MRI-based disease detection remains limited. In this work, we investigate two major self-supervised pretraining paradigms for MRI-based disease detection: reconstruction-based learning via Masked Autoencoders (MAE) and predictive representation learning via Joint Embedding Predictive Architectures (JEPA). We study the role of auxiliary objectives by introducing a novel spectral-domain reconstruction loss for MAE to enhance sensitivity to fine-grained anatomical structure, and by integrating variance--covariance regularization (VCR) within our JEPA framework to encourage decorrelated latent representations. Our models are pretrained on heterogeneous single-contrast MRI volumes in a contrast-agnostic setting, without modality concatenation. Across five downstream disease detection tasks, our results highlight the importance of self-supervised objective design for medical foundation model pretraining, demonstrating that the downstream benefit of each objective is determined by its relevance to the task's structure. Specifically, spectral regularization yields the largest improvements when the downstream discriminative signal is characterized by strong high-frequency anatomical structures, while covariance regularization is most beneficial when discriminative information spans multiple decorrelated feature dimensions. MAE with spectral-domain supervision consistently achieves superior downstream performance for MRI-based disease detection. These findings suggest that self-supervised objectives in medical imaging encode specific biases, and their downstream benefit is fundamentally conditioned on the task's structure.

The power distribution network is critical to reliable electricity delivery, yet traditional inspection methods face limitations in semantic understanding, generalization, and closed-loop automation. To address these challenges, this paper proposes a Multi-Modal Agent framework specifically for power distribution defect detection. Central to this study is the systematic evaluation of multimodal foundation models as unified cognitive engines. We rigorously assess their integrated performance across three critical capabilities: (1) Perception, where the model must accurately identify equipment and generate expert-level descriptions of defects; (2) Reasoning, where the model interprets visual findings to diagnose causes, assess severity, and plan maintenance strategies based on domain knowledge; and (3) Tool Usage, where the model acts as an autonomous operator to execute actions -- such as querying knowledge bases or generating work orders -- to achieve closed-loop maintenance. To support this evaluation, a domain-specific evaluation dataset and a comprehensive benchmark are developed. Experimental results demonstrate the strengths and limitations of current foundation models in these three dimensions, providing empirical evidence for deploying autonomous agents in high-stakes industrial environments.

Foundation models are rapidly transforming Earth observation by enabling scalable pretraining across diverse unlabeled geospatial modalities. However, their architectural diversity ranging from encoder-only to encoder-decoder and masked autoencoding paradigms makes it challenging to assess performance trade offs in a consistent manner. In this work, we present an apples-to-apples comparison of leading FM architectures designed for geospatial multimodal reasoning, with a particular focus on flexibility across varied spectral band configurations. We standardize pretraining using identical self supervised learning objectives and training datasets, and evaluate all models under consistent parameterization on the GEOBench benchmark across classification and segmentation tasks. Our results offer new insights into the design trade-offs between model flexibility, modality alignment, and downstream task performance. By highlighting architectural strengths and limitations under controlled conditions, this study provides practical guidance for building next generation geospatial foundation models capable of robust multimodal reasoning.

Decision-relevant building damage assessment is critical for prioritizing resources and recovery after a disaster, yet most automated methods either flatten damage into a single severity scale (no damage, minor, major, destroyed) or require paired pre- and post-event imagery that is often unavailable for emerging hazards. This paper presents Damage-TriageFormer, a single-image, post-event, footprint-conditioned model that produces a damage typology rather than a severity scale. We contribute: (1) DamageTriage-Bench, a new benchmark built from NOAA Emergency Response Imagery across Hurricane Michael (2018), Hurricane Helene (2024), and the 2025 Los Angeles wildfire complex, with five typology classes that distinguish roof damage from structural damage and, within each, partial from total extent; and (2) Damage-TriageFormer, which extends a DINOv3 ViT-L backbone with a Simple Feature Pyramid for higher-resolution instance pooling, a two-stage gated damage head, and an auxiliary severity-regression objective. Our model achieves macro F1 of 0.624 on validation and 0.619 on a held-out stratified test set, performing strongest where operational triage needs it most, with per-class F1 of 0.91 and 0.84 on undamaged buildings and total structural collapse, respectively. While the rare Total Roof Damage class remains difficult due to its limited examples and an inherently ambiguous label boundary, our results show that single-image post-event imagery can support actionable building damage typing, enabling targeted emergency response and resource allocation without a pre-event reference.

Generative AI emulators are increasingly used in scientific domains where we already have strong theory, benchmarks, and physical intuition. This raises a central evaluation and interpretability question: when a foundation-style model can reproduce known continuum dynamics, what internal mechanism supports that behavior, is the internal behaviour consistent with known physics, and how does it relate to where the emulator succeeds or fails? We investigate a cross-domain foundation model for continuum dynamics, Walrus by Polymathic, using mechanistic interpretability guided by physical principles. We apply a sparse autoencoder (SAE) to probe a selected layer, and address the practical challenge of triaging a large feature set (over 20,000) using enstrophy as a physically grounded metric. As a deliberately simple testbed, we focus on shear flow and compare feature recruitment across multiple shear-flow setups, i.e. parameter values in the numerical simulation. Across setups we find evidence of piecewise consistency, with subsets of features recurring in similar roles, but this structure is intermittent and does not map cleanly onto standard physical decompositions. In parallel, direct comparisons between numerical simulation and the emulator reveal systematic output-level discrepancies, including regimes where energy/structures become too diffuse or too localized. We connect parts of these discrepancies to changes in specific SAE feature usage. Our work highlights open questions for scientific foundation models: how to robustly prioritize mechanistically meaningful features, how to separate stable structure from analysis artifacts (including single-layer and SAE limitations), and how to use established benchmarks to decide when "different" internal representations are genuinely informative rather than merely effective.

Time-series foundation models (TSFMs) are increasingly explored as predictive experts within emerging agentic time-series systems. However, TSFMs exhibit heterogeneous inductive biases, and no single model consistently dominates across forecasting regimes, making expert selection a critical challenge. Existing systems often delegate this decision to LLM-based controllers, incurring substantial inference overhead. We present TimeRouter, an efficient routing framework that leverages empirical complementarity across a pool of pretrained TSFMs through lightweight discriminative routing, selective gating, and ensemble fallback. Concretely, TimeRouter combines a learned routing head, a selective gate, and an ensemble fallback, enabling adaptive expert selection without invoking an LLM at inference time. TimeRouter achieves state-of-the-art performance on the GIFT-EVAL leaderboard, with an LB MASE of 0.6765. Beyond benchmark performance, our ablation studies provide empirical insights into TSFM routing design, highlighting the importance of pool composition and selective gating. Taken together, these results position TimeRouter as a modular and lightweight routing layer for future agentic time-series systems built upon foundation-model pools. Our code is available at https://github.com/UConn-DSIS/TimeRouter.

Correcting handwritten exams by hand is time-consuming and error-prone, particularly for large cohorts, while fully digital exams tend to force a didactic narrowing towards closed question formats. A practical middle ground keeps paper-based, problem-oriented tasks but records the assessment-relevant answers as single capital letters in a table that a machine can read. The open question is whether this reading can be made accurate and, above all, fair enough for unsupervised grading. Earlier automated approaches reached only about 88%--91% recognition -- too low -- and failed on the cases that matter most: answers placed outside the cell, crossed out, or written in cursive. We show that general-purpose vision-language foundation models (VLMs), which interpret the page rather than match pixel templates, close this gap. On a benchmark of 61 anonymised exams (3141 answer positions) the best model reaches 98.4% accuracy, well above the previous baseline. Crucially, we centre the evaluation on fairness: we distinguish false negatives (a correct answer marked wrong, which disadvantages the student) from false positives, and a lightweight prompt that supplies the reference solution as context lowers the false-negative rate to 0.58%. Under an exemplary grading scheme only three of the 61 exams would be graded worse, all caught by a student self-review step. Fully automated, fairness-aware exam grading at scale is therefore defensible; we release the anonymised benchmark to support reproducibility.

Understanding and forecasting lake dynamics is critical for monitoring water quality and ecosystem health across lakes and reservoirs. While machine learning methods have been recently applied to ecological time-series data, existing works assume regular sampling in time and depth, and struggle to generalize across lakes with heterogeneous variables, depths, and observation patterns. To address these limitations, we introduce \textsc{LakeFM}, a foundation model for aquatic systems, pre-trained on large-scale ecological datasets comprising both simulated and observed lakes. Through extensive empirical evaluation, we show that \textsc{LakeFM} learns meaningful representations spanning broader lake-level characteristics, and achieves competitive or often superior-forecasting performance compared to existing time-series foundation and non-foundation models, while producing physically plausible predictions consistent with real-world lake dynamics.

Though wireless foundation models (WFMs) have shown strong potential in learning universal channel representations, their adaptation to various downstream tasks remains constrained by existing paradigms. Fine-tuning strategies introduces substantial computational and storage overhead, while frozen feature extraction leads to sub-optimal performance across diverse downstream tasks. To address this issue, we propose a unified adaptive feature composition framework for multitask generalization in WFMs, where the key component is the Routing Adapter for Feature Composition (RAFC). Instead of extracting only the final-layer output, this router treats the hidden states from different Transformer depths as a reusable pool of multi-level hidden features, and employs a lightweight task-driven feature composition network to generate layer-wise aggregation weights, then adaptively combine hierarchical representations through weighted summation. This design enables each downstream task to access suitable mixture of low-, mid-, and high-level wireless features without modifying the pretrained backbone. Extensive experiments on four representative wireless tasks demonstrate that RAFC consistently outperforms conventional adaptation baselines while introducing fewer than 50K additional parameters. Moreover, the learned routing weights provide interpretable evidence of task-specific layer preferences, making the proposed framework a low-complexity, scalable, and explainable interface for adapting WFMs to diverse downstream scenarios.

We study whether pretrained video foundation models encode intuitive-physics information in their frozen representations, and how this information varies across model families, layers, and probe types. Using frozen-feature probing on IntPhys2 and Minimal Video Pairs (MVP), we compare predictive joint-embedding models (V-JEPA), masked reconstruction models (VideoMAE), and a diffusion-based video generator (LTX-Video). V-JEPA achieves the strongest overall results across benchmarks, especially with probes that model temporal dynamics, while VideoMAE remains competitive and LTX-Video recovers weaker but non-trivial signal. Layerwise analyses show that physics-relevant information is weakest in early layers and becomes most accessible at intermediate-to-late depth, and temporal controls show that disrupting frame order substantially reduces performance, especially on MVP. Together, these results suggest that intuitive-physics knowledge emerges reliably in pretrained video representations, but its accessibility depends strongly on pretraining paradigm, representational depth, and readout mechanism.

Foundation models (FMs) are increasingly used as backbones for downstream tasks across language, vision, time-series, and multimodal applications. Yet existing model-serving systems deploy each customized task as an independent model instance, thereby replicating heavyweight backbones, wasting accelerator memory, and losing opportunities to amortize batching and loading costs. This paper presents FMplex, a serving system that treats FM backbones as a virtualization substrate for deployment sharing. FMplex presents each task with a virtual foundation model (vFM), a logically private FM instance backed by a shared physical FM. This abstraction lets independently customized tasks share a backbone while preserving task-specific extensions, independent lifecycles, and task-level isolation. In addition, we propose a batch-aware fair-queueing scheduler that combines weighted task-level sharing with inter- and intra-task batching across colocated tasks. We implement a FMplex-based serving stack spanning task construction, sharing-aware deployment, and runtime execution. Across 7 FM backbones (16 variants) and 92 downstream tasks, FMplex reduces latency by up to 80% over spatial partitioning and 33.3% over best-effort co-location, while hosting up to 6x more tasks at cluster scale.

Self-supervised data curation provides a pathway to scaling and improving the generalization capabilities of machine learning models. By leveraging self-supervised learning (SSL) for data curation, the demand for massive training datasets required by foundation models can be effectively met. SSL greatly alleviates the costs associated with annotation and manual dataset curation while minimizing the need for human oversight. However, the integrity of SSL-curated datasets must be rigorously checked, as reliance on anonymous and unvetted external sources can substantially increase the risk of data poisoning. In this paper, we propose a Poisoned Data Detector (PDD), an active defense mechanism designed to ensure the integrity of SSL-curated datasets prior to foundation model training. PDDs are designed using a combination of the pretrained ImageBind model and traditional classifiers, including Random Forest (RF), k-Nearest Neighbors (KNN), Naive Bayes (NB), and Support Vector Machines (SVM). We rigorously evaluated PDDs using 176,200 images from three diverse datasets and three different adversarial attacks encompassing both in-distribution and out-of-distribution scenarios. Notably, SVM-PDD achieves superior performance for both in-distribution (Set3-Set5) and out-of-distribution (TrueFace and 140K RealFace) datasets. Our design demonstrates strong scalability and enables the rapid integration of new adversarial attack detectors through an ensemble approach.

Remote-sensing foundation models (RSFMs) benefit from pretraining on imagery from multiple sensors and ground sampling distances (GSDs), but such exposure alone does not resolve scale mismatch during downstream adaptation. A fixed token-grid offset can correspond to different ground distances across sensors, making grid-based positional priors physically inconsistent. Meanwhile, heterogeneous spatial granularity means that compact urban regions and homogeneous landscapes may require different positional sensitivities even under the same GSD. Therefore, we propose {GeoRoPE}, a ground-aware, RoPE-compatible, and parameter-efficient spatial adaptation method for RSFMs. GeoRoPE recalibrates token-level positional interactions from two complementary aspects. First, \textit{Geo-Coordinate Calibration (GCC)} rescales raw token-grid offsets according to the ground distance represented by one token-grid step, producing geo-calibrated relative coordinates across GSDs. Second, \textit{Geo-Frequency Calibration (GFC)} adjusts the native RoPE frequency with a relation-specific factor, enabling position sensitive adaptation to scene-dependent spatial granularity. GeoRoPE is injected into pretrained RSFMs through a lightweight adapter, preserving the frozen spatial prior while adding geo-aware positional corrections. Experiments across multiple RSFMs, sensors, resolutions, and downstream tasks demonstrate that GeoRoPE improves cross-resolution robustness and scale-sensitive representation learning.

EEG foundation-model releases are usually audited one endpoint at a time: raw-reconstruction, membership inference, identity linkage, or DP-SGD on the downstream head. We audit the same released embeddings under all four endpoints jointly, on BIOT, LaBraM, and EEGPT, and show that each single-endpoint audit clears releases that still leak spectral attributes. The decisive evidence is a cross-encoder transfer audit: a single ridge attribute decoder learned from one frozen encoder transfers, via a fitted linear bridge, to held-out-subject test splits of every other encoder, with subject-disjoint matched-control 95% CI lower bound at least 0.081 across all six BIOT/LaBraM/EEGPT directions. We prove a sufficient condition: two encoders sharing a nontrivial attribute-coordinate projector overlap beta admit a chained ridge bridge attacker with centered-gain lower bound sqrt(beta/(1+tau^2)) - eps_br - rho_0, and back-solve beta in [0.008, 0.198]. To turn the joint audit into a deployment-readable decision rule we introduce an audit-endpoint disagreement score (AEDS), prove sufficient conditions for its positivity, and bootstrap-calibrate it per cell; AEDS is positive in all eight matched-CI cells (BIOT/LaBraM/EEGPT on EEGMMI; LaBraM on Sleep-EDF, 54-channel LIMO, CHB-MIT pediatric scalp EEG) with p<0.001, while a head-level Carlini LiRA membership audit reaches AUC only 0.50-0.70. Standard defenses fail under audit: a Wiener-style noise-aware adaptive attacker, the LiRA audit, and DP-SGD at every utility-preserving epsilon in {4,8} leave the attribute channel essentially unchanged. The contribution is an audit framework that turns scattered single-endpoint defenses into a joint release decision, supported by a cross-encoder bridge theorem and adaptive-attacker, LiRA, and DP-SGD baselines; the audit licenses release-blocking, not raw-waveform exfiltration or held-out-subject identity recovery.

Agentic AI systems can behave differently across runs: the same request may produce a different plan, a different tool call, a different code edit, or a different final answer. Such variability arises from several layers that are often conflated. At the core of many current agents is a foundation model, a large pretrained model adaptable to many downstream tasks, embedded in an orchestration loop that plans, calls tools, observes results, and updates state. One explicit intrinsic source of variability in such systems is token generation: the model computes scores over possible next tokens, the scores are converted into probabilities, and a decoder may sample tokens using a pseudo-random number generator. A small sampled token difference can then cascade downstream into a different tool call, code path, search query, or agent state. Other sources of variability are extrinsic to token sampling, including changing environments, live data, serving infrastructure, batch effects, and numerical details. By separating these layers, this tutorial clarifies what it means to call agentic AI systems stochastic, when such variability can be reproduced under matched conditions, and why deterministic execution need not imply identical behavior in deployed settings.

Multimodal Foundation Models (MFMs) have made substantial progress, yet remain fragile in spatial reasoning over the physical world. A key bottleneck lies in their inability to transform local egocentric observations into a global allocentric spatial representation. To address this, we propose AlloSpatial, an agentic framework for allocentric spatial cognition in foundation models. AlloSpatial introduces World2Mind, a plug-and-play cognitive mapping sandbox that converts egocentric observations into structured allocentric priors, including Allocentric-Spatial Trees and route maps that support querying object topology, geometric relations, passability, and trajectories. To utilize these priors reliably under noisy reconstruction and ambiguous visual evidence, AlloSpatial introduces a Spatial Reasoning Harness for tool-use judgment, modality-decoupled cue collection, and geometry-semantic arbitration. We further internalize this process in Qwen3-VL through cold-start reinforcement learning with a harness-gated trajectory-level reward. Experiments on VSI-Bench and MindCube show that AlloSpatial improves proprietary models by 5%-18% in a training-free setting, while ASTs alone support strong spatial reasoning even when visual inputs are removed. The trained AlloSpatial agents further outperform larger general-purpose models and competitive spatial baselines, suggesting that structured allocentric representations, active tool use, and verifiable reasoning offer a promising route toward spatially capable foundation models.

Test-time Scaling (TTS) has emerged as a pivotal research direction for enhancing model performance by dynamically allocating computational resources during inference. Recent advancements have adapted this paradigm to Multimodal Foundation Models (MFMs), unlocking their potential in multimodal reasoning and generation. Despite rapid progress, the field lacks a systematic survey and unified theoretical framework to delineate the developmental landscape of multimodal TTS. To bridge this gap, we present the first comprehensive review of TTS research for MFMs, proposing a unified taxonomic framework that categorizes existing methodologies into three distinct strategies: sampling-based, feedback-based, and search-based approaches. We further summarize representative applications and benchmarks commonly utilized to evaluate multimodal TTS capabilities in generation and reasoning tasks. Finally, this survey discusses open challenges and outlines future research directions, providing a systematic roadmap for subsequent studies in this rapidly evolving field.

Recent breakthroughs in self-supervised learning (SSL), such as the Latent-Euclidean Joint-Embedding Predictive Architecture (LeJEPA), alongside successes in integrating visual encoders with language models, have driven the demand for adaptable, high-capacity vision encoders in Computed Tomography (CT). In this work, we explore 2D slice-based architectures as a flexible alternative to native 3D models for processing volumetric CT data. Using the CT-RATE dataset, we trained DALE-CT (Depth-Aware Latent-Euclidean Computed Tomography), a 2D model family built entirely from scratch using LeJEPA, and compared its performance against a continually pre-trained DINOv2 baseline. To enhance representation quality, we developed a novel 3D depth-aware pre-training strategy anchored by dense auxiliary supervision from both automated anatomical masks and human-annotated abnormalities. Under linear probe evaluation with Multiple Instance Learning (MIL) for multi-abnormality detection, the frozen backbone of this dual-supervised model (DALE-CT-2S) achieves a Macro AUROC of 0.833. This performance demonstrates near-parity with state-of-the-art 3D vision-language models, achieved entirely from scratch with significantly less data and no textual supervision. To ensure reproducibility, all training code, evaluation scripts, and model weights have been made publicly available.

Foundation models for astronomical surveys offer powerful learned representations that can be transferred to downstream regression tasks such as galaxy property estimation. However, point predictions alone are insufficient for scientific inference; reliable uncertainty quantification (UQ) is essential. We compare seven UQ methods on galaxy property regression using frozen AION-1 foundation-model embeddings, predicting redshift, stellar mass, stellar-population age, gas-phase metallicity, and specific star-formation rate, from Legacy Survey photometry/imaging and DESI spectra, with PROVABGS-derived labels. Distribution-free conformal methods achieve marginal coverage within $\sim$1,pp of the nominal 90% across all properties, while non-conformal baselines (Deep Ensembles, MC~Dropout) fail to calibrate reliably. Among conformal approaches, Conformalized Quantile Regression (CQR) delivers the best coverage in the bin with the poorest model predictions. More importantly, only the Locally Valid and Discriminative (LVD) framework -- particularly when operating on AION-1 embeddings -- also provides finite-sample \emph{local validity}, producing intervals that adapt to each galaxy's local prediction difficulty rather than relying on marginal guarantees alone. These results establish conformal prediction, and LVD in particular, as the preferred UQ framework for uncertainty-aware inference on foundation-model embeddings in astrophysics.

At commissioning time, Photovoltaic (PV) operators must forecast production before target-site observations are available, limiting the direct use of standard supervised forecasters. This cold-start setting is addressed with a zero-shot pipeline that generates a synthetic production history from plant metadata and meteorological covariates, enabling time-series foundation models (TSFMs) to forecast through inference-time conditioning. Five TSFMs are benchmarked against classical baselines under strict Cold-Start Baseline, Real Feedback, and Self-Forecast Feedback strategies. The evaluation spans $440$ PV sites across four datasets and diverse climate regimes. Covariate-aware foundation models outperform baselines by approximately $1.7-2\times$: TabPFN-TS achieves the lowest error under Real Feedback (MAE $0.514$, RMSE $0.721$ $kWh$ ${kWp}^{-1}$ ${d}^{-1}$), while Chronos-2 is most robust under Self-Forecast Feedback. Performance is largely insensitive to the synthetic-history source, indicating that accuracy is driven more by the availability of plausible temporal context than by the specific generator.

Foundation models for wearable biosignals have matched or exceeded supervised specialists across a range of clinical tasks, yet all rely on modalities that require deliberate user action--wearing a device or visiting a sleep lab. We introduce BCG-FM, the first foundation model for ambient mechanical biosignals. A piezoelectric sensor embedded in the bed surface records ballistocardiography (BCG) each night without user effort; we pretrain BCG-FM with participant-level contrastive learning and using a total of 2.75 million hours of nightly recordings from 145,985 individuals, the largest raw-waveform biosignal pretraining corpus to date. Frozen BCG-FM embeddings achieve 3.26-year MAE on biological-age estimation (the lowest reported for any ambient, contactless modality) and yield clinically relevant discrimination across 15 self-reported health conditions and three independent external cohorts. Pretrained representations from only 500 labeled participants outperform a fully supervised baseline trained on 3,372, and representation quality scales log-linearly with contrastive batch size. These results establish ambient, longitudinal mechanical biosignals as a viable modality for health foundation models.

High-grade gliomas integrate into neural circuits through functional synapses with neurons, raising the question of which noncoding elements shape synaptogenic gene expression in tumor cells. The regulatory program written across the dark genome, what we call the $\textit{dark regulome}$, is the natural substrate to probe, and sequence foundation models offer a zero-shot route through in-silico mutagenesis (ISM); yet likelihood-based scoring is tautologically coupled to local sequence predictability, leaving the regulatory interpretation underdetermined. Across three architecturally distinct foundation models (Caduceus-Ph, HyenaDNA, Enformer) and 30,448 dark genome elements at 92 glioma-relevant loci, we introduce a residualization-and-permutation diagnostic that separates predictability-driven from regulation-driven RIS variance. A sharp 10kb proximal-regulatory horizon survives every control we apply, but the LM-derived element-class hierarchy does not: a six-feature linear baseline matches Caduceus top-decile membership at AUC $= 0.985$. Cross-architecture decomposition cleanly separates a sequence-predictability layer (the two language models co-rank long well-predicted transposable elements) from a regulatory-output layer (Enformer alone retains residual cCRE-discriminative signal), with literally zero overlap between the two top-100 lists. Conservation, brain cis-eQTL, and STRING-PPI cross-checks then anchor what biology survives: top-100 elements across all three models are $3.3\times$ enriched per model for matching brain eQTLs ($p_\mathrm{emp} < 5\times 10^{-3}$), while a tempting transposable-element regulatory layer and a striking NRXN1+NLGN1 protein-pair convergence both fail proper permutation tests once those tests are constructed. We deliver the diagnostic as a general methodological tool for any ISM-based regulatory study.

This paper presents a forward-looking vision for integrating the emerging multi-modal multi-task federated foundation models (M3T FedFMs) into vehicular networks, with the goal of unifying the expressive power of multi-modal multi-task foundation models (M3T FMs) with the privacy-preserving and distributed learning capabilities of federated learning (FL). Given the largely underexplored nature of this research direction, we first introduce the fundamental training/fine-tuning principles of M3T FedFMs. We then discuss a range of their representative use cases in vehicular networks, illustrating the significant potential of M3T FedFMs to enable next-generation vehicular intelligence. Afterwards, we identify key constraints inherent to vehicular environments that challenge the practical deployment of M3T FedFMs, and articulate a set of forward-looking research directions to address these challenges. Furthermore, through a case study conducted on a real-world vehicular dataset (i.e., Waymo Open Dataset), we demonstrate the promise of M3T FedFMs for vehicular networks and release our implementation to facilitate reproducibility and stimulate research in this emerging area (repository: https://github.com/KasraBorazjani/vehicular-fedfm)

Spatial transcriptomics (ST) is a powerful tool for exploring biological properties dependent on structure, proximity, and interaction in tissue. The methods underpinning ST are developing rapidly but are limited in their ability to profile many thousands of genes at a subcellular scale. Although dissociated from tissue, it is known that the whole-transcriptome readouts of cells in single-cell RNA sequencing (scRNA-seq) retain information about their former in situ neighbourhoods, motivating computational methods to recover it. While paired ST and scRNA-seq datasets are scarce, each modality in its own right is abundantly available. We therefore propose to perform cross-modal translation between unpaired ST and scRNA-seq data. In this work we show that a single-cell foundation model can perform this translation via adversarial fine-tuning. We demonstrate that our method performs favourably against methods built for multi-omics translation.

Wireless foundation models have emerged as a promising alternative to building separate models for each wireless task. However, existing approaches rely on masked input reconstruction, which can bias representations toward low-level signal details. In this paper, we propose LatentWave, a wireless foundation model pretrained using a Joint-Embedding Predictive Architecture (JEPA) on diverse wireless spectrograms and channel state information (CSI). By predicting masked regions in latent space, LatentWave learns representations that are more transferable out of the box across diverse downstream tasks. The proposed architecture employs per-channel patch embeddings with stochastic channel sampling during pretraining, allowing it to process variable antenna counts and improving usability across heterogeneous wireless configurations. We evaluate LatentWave on four downstream tasks: RF signal classification, 5G NR positioning, beam prediction, and LoS/NLoS classification, comparing against a masked-modeling baseline (WavesFM) pretrained on the same data. Additionally, we show that the masking geometry introduces a task-dependent inductive bias: frequency masking strongly favors channel-related tasks such as positioning and beam prediction, while region masking better preserves discriminability for signal classification.

This paper addresses the problem of attack detection in cyber-physical systems without any knowledge of the plant model or its structure. A remotely located plant transmits sensor measurements to an operator over a network that is assumed to be under attack. We consider two classes of attacks: model-free replay attacks and model-based stealthy attacks. For the latter, we derive closed-form expressions for the optimal stealthy attack policy against a $χ^2$ detector, for both linear and nonlinear systems. We then propose a model-structure-free detector based on TimesFM, a time-series foundation model developed by Google Research, which serves as a surrogate residual generator operating in a zero-shot fashion. We show empirically that the TimesFM-based detector achieves a comparable or superior attack detection performance. The efficacy of the proposed approach is demonstrated numerically on the IEEE 14-bus power system. We also demonstrate that TimesFM predictions can serve as a substitute for corrupted measurements, a practical mitigation technique when classical redundancy assumptions fail.

Entity alignment (EA) aims to identify equivalent entities across heterogeneous knowledge graphs (KGs) and is a key component of knowledge fusion and cross-KG reasoning. The recent EA foundation model demonstrates that alignment knowledge, once pretrained, can be directly applied to diverse previously unseen KG pairs. However, it still underuses structural context in two places: cross-KG interaction is weak during encoding, and final candidate ranking still relies too heavily on coarse similarity. We address these limitations with ContextEA, an enhanced encoder-decoder framework for transferable EA. On the encoder side, we introduce a cross-KG interaction encoder that unifies the two KGs with anchor bridges and performs earlier relation-aware cross-graph propagation. On the decoder side, we introduce a structural calibration decoder that calibrates alignment scores with entity-level, neighborhood-level, relation-level, and anchor-aware structural evidence. This design strengthens both structural context construction and structural context exploitation while remaining lightweight. Experiments on 29 EA datasets in OpenEA, SRPRS, and DBP show consistent gains over strong transferable baselines. Notably, the pretrained ContextEA already surpasses the finetuned baselines on all three benchmark groups, demonstrating substantially stronger transfer to unseen KGs. These results suggest that explicitly harnessing structural context is an effective direction for improving EA foundation models.

Controlled experiments are the backbone of machine learning research, but at the scale of modern foundation models, they have become prohibitively expensive. Instead, the community increasingly relies on research strategies that approximate the ideal experiment at a fraction of the cost: proxy experiments and scaling laws, observational studies with publicly available models, and single-run designs that leverage variation within individual training runs. In this work, we argue that there is no free lunch when approximating large-scale experiments on a compute budget. Specifically, savings in compute come at the cost of validity threats -- hidden and sometimes untestable assumptions that, when violated, can invalidate research claims. To help navigate such threats, we propose an evaluation framework that casts foundation model research as a causal inference problem. Within this framework, we evaluate different research strategies through four types of validity adapted from the empirical social sciences -- statistical, internal, external, and construct validity. We find that each strategy comes with a characteristic validity profile: proxy experiments trade external and construct validity for statistical and internal validity; observational studies face confounding and effect heterogeneity; and single-run designs are strained by interference between treated units. This analysis reveals several validity threats that have received insufficient attention in the literature. Overall, our evaluation framework provides researchers with a practical toolkit for scrutinizing validity threats in foundation model research~designs.

Adapting foundation models to detect generative artifacts via gradient-based updates compromises their intrinsic representations. Under optimization on limited samples, models overfit to local domain shortcuts. Fine-tuning massive weights on specialized data introduces erroneous inductive biases, inducing a measurable $\mathcal{L}_2$ norm perturbation in the high-dimensional feature space -- a phenomenon we formalize as anchor drift. Amplified by nonlinear activations, this drift impairs zero-shot forgery detection across unseen domains.We propose a gradient-free methodology reframing detection from binary classification to an out-of-distribution (OOD) anomaly measurement problem. Treating a frozen foundation model as a stable coordinate system, we establish an absolute natural anchor on the real visual manifold by analytically decoupling statistical and semantic deviations, derived from attention-weighted spatial moments and orthogonal projection of perceptual inconsistencies. Evaluated in an extreme zero-shot setting (trained on face forgeries, tested on universal Text-to-Image generations), our method significantly outperforms gradient-optimized paradigms. Backpropagation-free forward passes and linear solvers enable hardware-agnostic, edge-deployable calibration with minimal latency. Furthermore, the Sherman-Morrison formula unlocks instantaneous online learning against novel attacks and enables privacy-preserving federated collaboration via covariance delta transmission.

Whether attention maps from pathology foundation models capture genuine biology remains unknown, yet this question is critical for clinical trust and regulatory approval. We propose a spatial transcriptomics-based framework for orthogonal, hypothesis-free evaluation of attention and apply it to five pathology foundation models (CONCH v1.5, UNI v2, Virchow2, GigaPath, H-Optimus-1) and a ResNet50 baseline. Using attention-based multiple instance learning, we train single-task and multi-task models to predict five molecular alterations in glioblastoma on the CPTAC cohort, validate on an independent TCGA cohort, and evaluate biological coherence of attention maps against 87 transcriptional signatures using co-registered Visium spatial transcriptomics data from 18 samples. Internally, no single encoder dominates across all tasks, and external validation inverts internal performance rankings. Attention maps show a five-fold enrichment gradient from pathways (Cohen's d=0.329) to individual genes (d=0.055), indicating that attention captures emergent multi-gene transcriptional programs rather than individual molecular events. Spatially smooth attention maps do not imply biological coherence, and different encoders attend to distinct biological compartments. Our framework provides objective, quantitative assessment of what foundation models learn from histopathology, moving the field beyond qualitative saliency map review.

Tabular foundation models (TFMs) have made rapid progress in standard classification and regression, but time-to-event survival prediction tasks have remained largely untouched. Unlike in standard regression tasks, survival prediction models must account for censored data. Standard TFMs cannot handle natively censored data, leading to biased and inaccurate predictions, making them unsuitable for real-world applications. To overcome this fundamental limitation, we propose \texttt{SurvPFN}, a prior-data fitted network (PFN), for survival prediction tasks. We pretrain \texttt{SurvPFN} on millions of synthetic survival prediction tasks to learn survival via distributional regression that accounts for censored data. \texttt{SurvPFN} works by (1) generating data with Weibull event times and a non-informative censoring mechanism; (2) integrating a censored event indicator; and (3) minimizing a censored negative log-likelihood. On SurvSet, a collection of real-world survival tasks, \texttt{SurvPFN} is highly competitive with classical and deep survival baselines without per-dataset fitting, a survival-specific architecture, or feature engineering. We show that survival can be treated as a continuous-time distributional regression problem with censored loss, unlocking the power of PFNs for time-to-event predictions.

Previous work has evaluated physics reasoning in foundation models using synthetic or semi-synthetic scenes and visual question-answering tasks. However, these benchmarks emphasize high-level events and lack the visual fidelity required to assess true low-level Newtonian understanding. We introduce NewtPhys, a 4D physically annotated dataset built from multiview images of real-world scenes with physics-grounded simulations. The dataset provides dense, fine-grained annotations across timesteps -- including 3D forces and amodal per-pixel quantities covering physics, tracking, semantics and geometry -- bridging the gap between simplistic synthetic setups and realistic visual complexity. Using NewtPhys, we systematically evaluate 56 VLMs, including 54 open-weight models and 2 closed-source frontier models, and 10 VFMs and reveal limitations in low-level physics reasoning. Beyond benchmarking, our dataset enables future research in physics-grounded vision and the development of next-generation physics-aware evaluations. Code and datasets are available at https://astra-vision.github.io/NewtPhys.

Existing human attention (saliency) modeling methods persist as highly fragmented across modalities, scenes, and task formulations. Consequently, even with increasing model capacity and data scale, current models predominantly remain scene-dependent and task-specific, failing to practically generalize in real-world applications. To address the fundamental limitations, we present the Attend to Anything Model (AAM), a multi-modal foundation model that unifies attention modeling across various image, video, and audio-visual tasks and scenes. AAM reformulates attention as a cognitive entailment relationship organized in a general-to-specific hierarchy, implemented through language prompts with hierarchical embeddings in hyperbolic space. Furthermore, to unify static image and dynamic video attention, we adopt a fluid-dynamics perspective, formulating video-frame attention as a diffusive temporal evolution governed by the Fokker--Planck equation. Extensive experiments on 16 benchmarks demonstrate that AAM consistently outperforms state-of-the-art methods by an average of 6% across various scenarios, while achieving approximately a 4$\times$ speedup in video inference. Overall, these results demonstrate that AAM provides a principled foundation for future research on attention and saliency-related tasks. The dataset and code will be available at https://github.com/wz-zhao/Attend-to-Anything.

Parameter-efficient fine-tuning (PEFT) is usually treated as a cheaper alternative to full fine-tuning. We study a broader role: small trainable adapters as persistent local state on top of strong shared foundation models. In this framing, the base model provides shared competence while adapters carry instance-specific behavior such as preferences, skills, tool habits, and memory-like updates. We organize the problem around three scaling axes: Scale Up, where stronger shared priors make small local updates more useful; Scale Down, where we study how small adapters can be while remaining reliable; and Scale Out, where many persistent adapted instances coexist. MinT provides one infrastructure example for managing adapter identity, revision, provenance, evaluation, and serving residency. Together, the results suggest that PEFT can be a compact substrate for persistent personal models rather than only a budget substitute for full fine-tuning.

This paper proposes ``FlatVPR,'' a novel geometric rectification paradigm that effectively bridges the trade-off between map lightweightness and localization accuracy in visual place recognition (VPR) by enforcing a feature manifold structure where any descriptor between two adjacent anchors $\mathbf{z}_A$ and $\mathbf{z}_B$ can be accurately reconstructed via linear interpolation $\hat{\mathbf{z}}_{pseudo} = (1-t)\mathbf{z}_A + t\mathbf{z}_B$, where $t \in [0,1]$ denotes the relative position. While state-of-the-art foundation models such as DINOv2-ViT-S/14 provide robust semantic features, their latent manifolds exhibit prominent curvature, projecting uniform linear motion in physical space onto highly non-linear trajectories in the feature space, which hinders reliable reconstruction under sparse anchor conditions. To enable the aforementioned interpolation-based reconstruction, we introduce a residual transformation $\hat{\mathbf{z}} = \mathbf{z} + \text{Res}(\mathbf{z})$ to the raw foundation features $\mathbf{z}$, where $\text{Res}(\cdot)$ represents a learnable adapter. Our method explicitly suppresses manifold curvature using a mathematically grounded Pullback Flatness Loss that minimizes the deviation of intermediate features from the linear segment connecting adjacent anchors, thereby minimizing the intrinsic curvature of the manifold. Through this spatial flattening, map construction is formulated within an Expectation-Maximization (EM) framework, decoupled into a continuous M-step for manifold adaptation and a conceptual E-step for optimal anchor selection guidelines. Experiments on the NCLT dataset demonstrate that the application of our adapter leads to significant performance improvements even under extremely sparse anchor conditions with 100m intervals and extreme seasonal changes.

Foundation models (FMs) have achieved substantial success in generalizing across tasks without problemspecific training or fine-tuning. However, many critical applications in mechanics and computational science require not only accurate predictions but also reliable uncertainty quantification (UQ). Herein we investigate the UQ capabilities of tabular FMs in regression tasks through a comprehensive empirical study comparing Tabular Prior-Data Fitted Networks (TabPFN) against Gaussian processes (GPs). We systematically evaluate these two methods across a host of regression problems with varying complexity, dataset sizes, and input dimensionalities. We use a default setting to build all the GPs and for a fair comparison against TabPFN v2.5. Our findings highlight an important trade-off between explicit and learned priors: while TabPFN achieves highly competitive performance for complex, high-dimensional problems with sufficient data, GPs often provide superior predictive accuracy and UQ in data-scarce settings. Moreover, when the chosen kernel constitutes a good prior for the underlying function, GP performance can substantially exceed that of TabPFN. Our results can be reproduced from https://github.com/kianswarehouse/GPvsPFN.

Humans can reproduce the viewpoint specified by a target image through active head and body motion, yet spatial intelligence in foundation models has largely been studied as passive understanding of pre-collected observations. We introduce Target Viewpoint Reproduction (TVR) -- an active task where an agent adjusts its viewpoint in a 3D environment until its observation matches a given target image -- and TVRBench, an indoor-simulation benchmark spanning scene scale and target-view visual richness. TVR is far from solved: on the evaluation split, the strongest open-source and closed-source models reach only 7.8% and 12.0% success. Fine-grained analysis identifies two consistent bottlenecks: off-the-shelf models struggle with multi-turn visual history, and performance drops sharply when viewpoint reproduction requires body translation rather than in-place rotation, exposing a gap in mapping spatial discrepancies to embodied movement. To study reducing this gap, we build a unified TVR post-training framework covering expert-trajectory SFT, rationale-supervised CoT-SFT, offline Single-turn GRPO, and on-policy Multi-turn GRPO from live simulator rollouts. Visual-action SFT supplies the main gain, raising a 9B open-source model to 50.8% success; Multi-turn GRPO provides targeted multi-room refinement and reaches 51.4% overall, while CoT supervision and Single-turn GRPO degrade closed-loop performance. These results establish TVRBench as a testbed for measuring and training foundation models that actively perceive and act in 3D environments. Our code, data, and models are available at https://github.com/aim-uofa/TVRBench.

Multiplexed fluorescence microscopy improves tissue segmentation by providing complementary channels including nuclear (DAPI) and membrane (E-cadherin), that together encode richer spatial context than single-channel imaging alone. However, multiplexed models require all channels at inference, limiting deployment where only a subset is available. This work proposes a cross-modal knowledge distillation framework that transfers semantic information from a frozen foundation model teacher processing multiplexed input to a lightweight student operating on the nuclear channel only. The distillation objective combines MSE-based probability matching, boundary-aware supervision, and learnable uncertainty weighting. SAM ViT-H and CellSAM are evaluated as teachers across four U-Net students: Swin-Tiny (27M), ResNet18 (11M), EfficientNet-B0 (5.3M), and MobileNetV3 (1.5M), on TissueNet and BBBC038. On TissueNet, the SAM-distilled Swin-Tiny student achieves Dice 78.36 (plus or minus 1.44), a 13.05-point improvement over the no-KD baseline (65.31 plus or minus 1.35) and 87.9% recovery of teacher oracle performance (89.12 plus or minus 1.21) at a 23x parameter reduction. KD consistently improves all four students by approximately 12 Dice points, confirming architecture-agnostic distillation. SAM ViT-H outperforms CellSAM as teacher across all settings. Cross-dataset evaluation on BBBC038 shows consistent gains without teacher retraining.

Measuring statistical dependency between high-dimensional random variables is a fundamental task in data science and machine learning. Neural mutual information (MI) estimators offer a promising avenue, but they typically require costly iterative optimization for each new dataset, making them impractical for real-time applications. We present InfoAtlas, a foundation model-like architecture that eliminates this bottleneck by directly inferring MI in a single forward pass. Pretrained on large-scale synthetic data with rich dependence patterns, InfoAtlas learns to identify diverse dependence structures and predict MI directly from the dataset. Comprehensive experiments demonstrate that InfoAtlas matches state-of-the-art neural estimators in accuracy while achieving $100\times$ speedup, can flexibly handle varying dimensions and sample sizes through a single unified model, and generalizes effectively to complex, real-world scenarios. By reformulating MI estimation as an inference task, InfoAtlas establishes a foundation for real-time dependency analysis.

This study introduces a novel Arctic-focused remote sensing foundation model (RSFM) by combining diversity-aware regional-scale image curation with masked autoencoder (MAE) self-supervised pretraining of a Vision Transformer (ViT) encoder for very-high-spatial-resolution (VHSR) satellite image analysis. Spectral and acquisition-metadata descriptors were used in a scalable affinity-propagation clustering workflow to select approximately 3 million chips from 267 TB of Vantor VHSR imagery This curation strategy was designed to reduce oversampling of visually repetitive or low-information areas while preserving broad scene diversity across the study domain. We pretrained a ViT-Large encoder on the curated corpus using a domain-adapted MAE reconstruction objective, producing Arctic-specific transformer weights for downstream feature mapping. The pretrained encoder was integrated into an existing location-aware detection and segmentation framework and evaluated across four hand-labeled Arctic datasets. Compared to ImageNet-initialized ViT-Large baseline, Arctic MAE pretraining produced consistent improvements in foreground mean F1 scores of 0.87, 0.72, 0.93, and 0.87, for infrastructure, IWP, RTS, and TCNs, with approximately 5-8 percentage increase. The proposed model also outperformed Prithvi-EO-2.0 in all downstream comparisons, with the smallest gain corresponding to at least a 15 percentage improvement mean F1, suggesting that domain-specific self-supervised pretraining on curated Arctic VHSR imagery provides more transferable representations for fine-scale Arctic mapping than a general-purpose Earth observation foundation model. These results demonstrate that optimizing the pretraining data distribution at regional scale, while keeping the architecture and MAE objective fixed, can produce a reusable Arctic-domain encoder for multiple VHSR remote sensing applications.

Public numeric benchmarks appear in pretraining, so an evaluation that conditions on a date may be measuring memorized recall rather than out-of-sample skill. We introduce NumLeak, a measurement framework that combines API-boundary probes on production models with a white-box controlled validation on an open causal LM. Top-tier frontier LLMs recall the Fama-French market excess return at 3-seed pooled Pearson r=0.97-0.99 while staying within 0.15 within-25bps on the five sibling factors; comparable fidelity appears on U.S. unemployment, CPI inflation, and NOAA temperature. On a recent-release holdout, parse rate collapses to 21-57% but r stays at approximately 0.99 on months answered, the refuse-or-recall asymmetry a memorized channel predicts. The white-box experiment reproduces the dose-response, and logprob ranking detects memorization that open-ended generation misses, implying closed-API black-box probes understate the channel. A Sonnet "date to market-sentiment" regression that correlates with true Mkt-RF at r=0.74 collapses to r=0.02 once the model's own recall is residualized out. A one-line system-prompt defense blocks 99.8% of a non-adaptive single-turn suffix attack set at near-zero utility cost on conceptual and historical-narrative queries

Graph foundation models (GFMs) aim to reuse a single backbone across diverse graph domains, yet their transfer is often uneven and can exhibit negative transfer. While most prior work improves transfer through architectural or adaptation choices, we ask a data-centric question: which properties of two graph domains determine how much a fixed representation model changes its outputs? Using a graphon-based continuous limit for dense graphs, we show that for both set-based and message-passing tokenizations, any Lipschitz backbone admits an explicit decomposition of cross-domain output shift into (i) graph-specific finite-sample approximation terms and (ii) an intrinsic, relabeling-invariant domain discrepancy capturing structural mismatch. A key ingredient is positional-encoding (PE) stability: we establish stability guarantees for spectral PEs and highlight contrasting behaviors of eigenvector- versus subspace-based PEs. Experiments on synthetic and real graphs validate the theory and translate the decomposition into guidance for data curation in GFM transfer.

Electroencephalography (EEG) is a widely used non-invasive technique for measuring brain activity in brain-computer interface (BCI) applications. Supervised EEG decoding models often struggle to generalize across tasks, subjects, and datasets, motivating transformer-based EEG foundation models trained with self-supervised learning. Since transformers are permutation-invariant, they require explicit positional information. Unlike textual tokens, EEG electrodes are spatially distributed across the scalp, raising the question of how electrode positions should be encoded in transformer-based EEG models. In this study, we benchmark five positional encoding strategies within the CBraMod backbone and evaluate them under linear probing and fine-tuning protocols on motor imagery classification and emotion recognition. Our results show that no single strategy consistently outperforms across tasks. Spherical Positional Encoding (SPE) yields strong representations for motor imagery but underperforms on emotion recognition, while Asymmetric Conditional Positional Encoding (ACPE) demonstrates more consistent performance across tasks. These findings suggest that the optimal positional encoding strategy is task-dependent, with no universal solution across EEG decoding scenarios.

Federated fine-tuning of foundation models with Low-Rank Adaptation (LoRA) provides an efficient solution for reducing communication and computation costs while preserving data locality. However, the direct combination of FedAvg and LoRA suffers from three key issues: limited update space, which restricts the model's effective learning capacity; inter-round state mismatch, which disrupts cross-round local optimization continuity; and a client-agnostic starting state, which slows local convergence on clients. Although recent methods mitigate the limited update space issue by merging LoRA updates into the backbone across communication rounds, inter-round state mismatch and the client-agnostic starting state remain insufficiently addressed. To address these issues, we propose FedSmoothLoRA, a federated LoRA tuning framework that preserves the enlarged update space, improves cross-round local optimization continuity, and provides a client-aware starting state for local training. At each communication round, FedSmoothLoRA constructs the local LoRA initialization using two matrices: a Round-Matching matrix that preserves cross-round local state continuity, and a Gradient-Aligned matrix that provides client-specific optimization guidance from gradient signals estimated on local data. Together, these designs enable smoother and faster convergence. Extensive experiments on image classification and natural language generation tasks demonstrate that FedSmoothLoRA consistently outperforms existing federated LoRA tuning methods. Code: https://github.com/wangzehao0704/FedSmoothLoRA

Knowledge distillation (KD) transfers a single scalar prediction from a large foundation model (FM) to compact vertical models (VMs), suffering from diminishing transfer ratio -- the fraction of FM improvement captured by the VM -- as a single scalar cannot convey the rich intermediate knowledge that larger FMs learn. To address this bottleneck, we propose LoopFM (Learning frOm HistOrical RePresentations of FM), a framework that opens a high-bandwidth transfer channel by structuring FM intermediate embeddings as input features (e.g., user history sequence) for downstream VMs, without requiring real-time FM inference at serving and architectural coupling between FM and VM. We provide a theoretical framework for LoopFM with a gain decomposition and transfer-ratio analysis. On three public benchmarks, LoopFM demonstrates strong AUC improvements (e.g., 6%+ on TaobaoAd) and complementary knowledge transfer capability with KD. On industrial-scale systems (billions of examples, trillion-parameter FMs), LoopFM approximately doubles the knowledge transfer ratio on top of KD, delivering a +0.5% conversion improvement in the first half after its initial launch, and +1.03% and +1.22% conversion improvement from two individual launches in the subsequent half.

Ultrasound is widely used in obstetric care due to its safety, accessibility, and real-time imaging. However, interpretation remains operator-dependent and susceptible to noise and artifacts. Deep learning models have shown strong performance to solve these problem, but they typically require large annotated datasets that are difficult to obtain in clinical ultrasound. Foundation models (FMs) offer an alternative, using a large number of ultrasound images to learn transferable representations that can generalize with limited labeled data. This work presents a comprehensive benchmark of ultrasound-specific FMs for fetal plane classification. We evaluated four ultrasound FMs (USFM, MOFO, UltraSAM, FetalCLIP) against two CNN baselines (ResNet50, EfficientNet-V2) and a ViT (DINOv3) pretrained on natural images. We trained all models under two complementary settings: full fine-tuning and linear probing with a frozen encoder. All models were trained using 5-fold patient-level cross-validation on a Spanish fetal ultrasound dataset and tested on both in-domain data and an external African cohort to assess cross-population generalization. We found that FetalCLIP achieved the best results in the linear probing setting (F1 = 0.9261 for in-domain, F1 = 0.9731 for out-of-domain), while USFM performed best in the full fine-tuning setting (F1 = 0.9476 for in-domain, F1 = 0.9515 for out-of-domain). MOFO and UltraSAM degraded most in both settings, underperforming natural image pretrained models in some cases. These findings highlight how the choice of pretrained model strongly affects fetal plane classification performance, since different pretraining objectives lead to different levels of transferability.

While spatial foundation models have demonstrated impressive performance on standard datasets, a critical question remains: are they truly all-round players capable of generalizing robustly across diverse downstream tasks, arbitrary viewpoints, shifting scene domains, varying input densities, and specific hardware constraints? Answering this overarching question requires a holistic assessment, yet current models are mainly evaluated on specific domains for which they were specifically designed or trained. Such evaluations are intrinsically limited by narrow paradigm coverage, limited scene domains, and arbitrary frame sampling, making it fundamentally difficult to assess their true generalization capabilities. To address this gap, we present SpatialBench, a cross-paradigm, domain-diverse benchmark for spatial foundation models with deterministic sampling. SpatialBench features unprecedented scale and rigorous deterministic design, comprising 19 datasets and 546 scenes across 5 diverse spatial domains. It comprehensively evaluates 41 models across 6 paradigms on 5 task suites under 4 different input density settings. Our extensive evaluation reveals that current models are not yet all-round players, and uncovers crucial insights for future advancement. Specifically, we demonstrate that full-context attention maximizes accuracy while bounded-memory strategies unlock long-sequence scalability. Moreover, our empirical evaluations in challenging embodied and egocentric tasks demonstrate that strict domain alignment and high data quality are far more critical to performance than simple dataset scaling. Furthermore, to address the largest data gap identified in our analysis, we go beyond evaluation by introducing a large-scale dataset, DA-Next-5M, and a strong baseline model, DA-Next, pushing the boundaries of spatial representation learning.

Selecting which instances to label is a key challenge in low-label tabular learning. For recent Tabular Foundation Models such as TabPFN, context selection directly determines predictive performance. Supervised oracle experiments show that carefully chosen labeled context sets can strongly outperform random selection under the same labeling budget. However, the cold-start setting, where instances must be selected before any labels are available, has received little attention in the TFM literature. This problem is fundamentally geometric. In vision and language, foundation models induce embedding spaces where simple geometric selection methods are effective. In contrast, tabular instance selection has so far been performed predominantly in the original tabular space, which lacks a natural metric; heterogeneous types, mixed scales, and nonlinear interactions make raw-space distances unreliable for context construction, and original-space selection falls below random on the majority of datasets as the budget grows. We propose LUCoS (Latent Unsupervised Context Selection), which replaces raw-feature geometry with the latent geometry induced by embeddings from an unsupervised Prior-Fitted Network (PFN) and selects representative medoids as context. Evaluated on 67 OpenML-CC18 datasets across six low-label budgets, LUCoS ranks first under mean AUC, ACC, and F1, with conclusions stable across metrics and dataset-level robustness checks. A gain decomposition reveals a simple mechanism: at the smallest budgets, the main benefit comes from enforcing coverage; as the budget increases, the decisive factor becomes the representation space in which coverage is measured. LUCoS mitigates failures of original feature space selection, showing that reliable unsupervised context selection depends less on selector sophistication than on defining representativeness in a meaningful representation geometry.

We address the challenging task of 3D object segmentation in complex scene point clouds without relying on any scene-level human annotations during training. Existing methods are typically constrained to identifying simple objects, primarily due to insufficient object priors in the learning process. In this paper, we present FoundObj, a novel framework featuring a superpoint-based object discovery agent that incrementally merges suitable neighboring superpoints, guided by our innovative semantic and geometric reward modules. These modules synergistically leverage semantic and geometric priors from self-supervised 2D/3D foundation models, providing complementary feedback to the object discovery agent and enabling robust identification of multi-class objects through reinforcement learning. Extensive experiments on diverse benchmarks demonstrate that our approach consistently outperforms existing baselines. Notably, our method exhibits strong generalization in zero-shot and long-tail scenarios, underscoring its potential for scalable, label-free 3D object segmentation.

Childhood anemia affects around 40% of children aged 6-59 months globally and arises from heterogeneous factors, limiting model generalizability. We evaluate a transformer-based tabular foundation model against classical supervised methods under cross-country and data-scarce settings. We used DHS data from 16 countries across Africa, Asia, Latin America, the Caucasus, and the Middle East (n=68,856). We compared Logistic Regression, XGBoost, LightGBM, and TabPFN v2.6. Performance was assessed using AUC-ROC, Brier score, and ECE. Generalization was evaluated using leave-one-country-out (LOCO), reverse-LOCO, and few-shot settings. Subgroup analyses included sex, age, residence, maternal education, and wealth. Feature importance was estimated using SHAP. TabPFN outperformed classical models in low-data regimes (<200 samples), showing higher discrimination and better calibration. Across countries, it achieved the lowest Brier score (0.042) and ECE (0.203). Under full-data settings, AUC-ROC ranged from 0.59-0.76 with small between-model differences ($\leq 0.05$). LOCO performance was stable (0.58-0.69), driven by country context. Reverse-LOCO showed asymmetric transferability. Subgroup performance was consistent with no systematic demographic bias. SHAP identified child age, altitude, and height-for-age z-score as dominant predictors, followed by wealth and maternal education. Performance in childhood anemia prediction is driven more by population variation than model choice. TabPFN provides advantages in low-resource settings through improved discrimination and calibration, highlighting foundation models as promising tools for data-scarce global health prediction.

Tabular foundation models achieve strong accuracy on choice prediction tasks, but their predictions often violate the economic logic those tasks require: raising a price sometimes increases predicted demand, and implied willingness-to-pay estimates are frequently negative or implausible. We propose a two-stage adapter that embeds foundation model predictions within a utility-maximization framework. In the first stage, we estimate a standard choice model whose parameters are constrained to obey economic theory. In the second stage, we freeze those parameters and train a correction term that incorporates the foundation model's predictions as additional information. The result is a model that inherits the foundation model's accuracy gains while guaranteeing monotonic price-demand relationships under policy perturbation and producing analytically computable trade-off measures. On two transportation datasets, the adapter recovers up to 13 percentage points of accuracy over a standard logit model while maintaining perfect economic consistency, something neither the raw foundation models nor conventional distillation achieve.

Geometry estimation from perspective images has greatly advanced, maturing to the point where off-the-shelf foundation models are able to reconstruct 3D scene structure not only from multi-view imagery, but even from a single view. A natural extension is 3D reconstruction from panoramas, with the exciting prospect of recovering a full 360-degree scene from a single panoramic image. In this work, we introduce PaGeR (Panoramic Geometry Reconstruction), a framework to lift powerful 3D foundation models designed for perspective imagery to the panorama domain. Our strategy is to start from a pre-trained transformer for 3D reconstruction and turn it into a unified high-performance model that predicts scale-invariant depth, metric depth, surface normals, and sky masks from both perspective and omnidirectional images, in a single forward pass. By keeping architectural changes to a minimum and mixing perspective and panoramic images during training, PaGeR retains the rich 3D prior of the underlying foundation model while learning to also estimate geometrically consistent 360-degree scenes from single panoramas. We extensively test our method in both indoor and outdoor environments and find that it delivers state-of-the-art performance and excellent zero-shot performance across a wide range of scenes. Code, data and models are available $\href{https://github.com/prs-eth/PaGeR}{\text{here}}$.

This paper studies the next major bottleneck in agentic AI as system scaling, not only model scaling: the design of auditable, persistent, modular, and verifiable architectures around foundation models. We refer to this shift as scaling the harness: treating the structured execution layer around a foundation model as a first-class object of design, evaluation, and optimization. Although recent large language models enable agents to use tools, retrieve information, maintain memory, and execute long-horizon workflows, evaluation remains largely model-centric, often reducing agents to final-task success while treating memory, retrieval, tool use, orchestration, verification, and governance as secondary implementation details. This framing is increasingly inadequate because agent performance emerges from the interaction among the foundation model, memory substrate, context constructor, skill-routing layer, orchestration loop, and verification-and-governance layer. Together, these components form the agent harness, which translates model capability into long-horizon agent behavior. We study scaling the harness through three core bottlenecks: context governance, trustworthy memory, and dynamic skill routing, together with the orchestration and governance mechanisms that coordinate and constrain them. We further outline a research agenda for harness-level benchmarks that go beyond one-shot task success to measure trajectory quality, memory hygiene, context efficiency, communication fidelity, verification cost, and safe evolution over time. To make the discussion concrete, we develop CheetahClaws: https://github.com/SafeRL-Lab/cheetahclaws, a Python-native reference harness, and compare it with Claude Code and OpenClaw. Our main claim is that future progress in agentic AI will depend as much on system design as on stronger foundation models.

Light sheet fluorescence microscopy (LSM) enables high-resolution, three-dimensional (3D) imaging of biological specimens, providing rich volumetric data for studying cellular organization, pathology, and vascular networks. However, the size, dimensionality, and annotation burden of LSM data make supervised deep learning approaches costly and difficult to scale. Additionally, despite the abundance of unannotated LSM volumes, foundation models for this modality remain underexplored due to computational challenges and the complexity of volumetric representation learning. In this work, we introduce a 3D foundation model for LSM data, pretrained on a large curated collection of 3D images spanning multiple organisms, stains, and imaging protocols. We learn transferable volumetric representations by jointly optimizing for masked reconstruction and image-text alignment. The pretrained backbone drastically reduces the annotation burden, enabling efficient, few-shot adaptation for varied downstream tasks. We evaluate this approach on downstream segmentation, classification, and deblurring. Our results demonstrate consistent improvements over baselines, (1) when measured using standard evaluation metrics and (2) when rigorously assessed by domain experts. This highlights the potential of foundation model pretraining to reduce annotation requirements while improving performance across diverse LSM analysis tasks. Pretrained model weights and code for pretraining and finetuning are publicly available: https://github.com/AdinaScheinfeld/lsm_fm_public_repo.git.

Robotic perception in unstructured environments remains challenging despite the zero-shot capabilities of foundation models such as SAM. This work attributes performance degradation to non-uniform representation shifts across transformer layers: shallow layers exhibit substantial domain gaps (CKA < 0.5), whereas deep layers transfer effectively (CKA > 0.7). Based on this observation, we propose RepSAM, a representation-guided parameter-efficient fine-tuning (PEFT) framework for adapting foundation models to robotic vision. RepSAM employs a theoretically grounded CKA-guided rank allocation strategy combined with a multi-modal fusion module for robust handling of challenging robotic scenarios, including transparent objects and cluttered scenes. Experimental evaluation across six benchmarks and robotic manipulation tasks demonstrates that RepSAM achieves 97.9% of full fine-tuning performance (89.0% vs. 90.9% mIoU) while reducing trainable parameters by 158x (from 632M to 4.0M). RepSAM outperforms DoRA by 7.9% mIoU with just 4 hours of training on a single A100 GPU (a 96x reduction from full fine-tuning, which takes 384 GPU-hours). These improvements are statistically significant (p < 0.01) and translate to a 12.0% absolute improvement in robotic manipulation success rates over the LoRA (RGB) baseline.

Competitive resource allocation problems over frequency and space can be formulated as minimax interaction between transmit power and worst-case interference. This formulation naturally arises in multi-operator low Earth orbit (LEO) satellite spectrum sharing, where transmissions from competing constellations interfere in real-time. Under Gaussian channels, the corresponding power-allocation problem admits a convex-concave formulation with a unique saddle point. Discrete constellations yield generally nonconvex mercury/water-filling formulations. In this paper we propose the adversarial water-filling (AWF) problem with corresponding theory and algorithms for these settings. In addition, we develop a domain-specific wireless foundation model for AWF to learn the AWF search dynamics. The architecture incorporates permutation-invariant channel representations, a constraint-aware graph neural network (GNN) with sparse message passing, and global latent variables capturing the low-dimensional water level implied by the AWF optimality. Through learned projected extragradient iterations, the model approximates stationary solutions of the constrained minimax problem arising under mercury/water-filling. We further establish projected-stationarity/Karush-Kuhn-Tucker consistency and conditional local convergence of the learned AWF dynamics under local regularity and stability conditions. Experiments demonstrate empirical generalization across unseen problem sizes, constraint structures, discrete constellations, and channel-weighted objectives, while achieving a median speedup exceeding one order of magnitude over Mirror-Prox on matched instances at comparable first-order solution quality. The related code can be found at https://github.com/convexsoft/AWF.

Time series foundation models (TSFMs) are increasingly pretrained on large corpora, raising concerns that evaluation datasets may have been exposed during pretraining and thus yield overly optimistic performance estimates. Auditing such contamination is challenging in time series because signals are continuous and heterogeneous, and often lack corpus documentation. To the best of our knowledge, this is the first work to study pretraining contamination auditing for TSFMs. We formalize the problem of pretraining contamination auditing for TSFMs and propose TSFMAudit, a method based on probe adaptation dynamics. Our key intuition is that contamination manifests as unusually efficient adaptation: after a fine tuning probe, contaminated datasets tend to exhibit faster loss reduction with smaller backbone movement. We evaluate TSFMAudit on 6 TSFMs and 187 datasets using documented training source evidence as supervision, and compare against 10 competitive baselines adapted from the LLM literature.

Foundation models can increasingly describe, reconstruct, and generate 3D objects, assemblies, scenes, and environments, but visually plausible spatial output is not yet operable 3D. A generated object or environment becomes useful to an agent only when the system can identify its entities, frames, surfaces, constraints, provenance, admissible actions, expected effects, and validation failures. This paper introduces Hylos, a systems architecture for contract-bounded spatial intelligence. Hylos maintains scene-scale operability state over objects, assemblies, assets, surface anchors, assertions, action candidates, solver jobs, shared actuator invocations, capability gaps, and effect diffs. Durable spatial changes are routed through a SpatialTransaction: a commit boundary that resolves references, checks admissibility, protects invariants, projects effects, and returns commit, review, rollback, deferral, or capability-gap outcomes. The paper is framed as a systems/position preprint with a focused artifact study rather than a broad benchmark. The study examines causal repair: a visible misalignment appears on a dependent component, while the supported repair lies upstream in the placement structure that controls it. The successful interaction traces the symptom through scene dependencies, selects a supported upstream interaction, and applies a validated change instead of directly editing visible geometry. The broader claim is that spatial AI should be evaluated not only by visual quality, but by whether generated or edited 3D can become reliable substrate for CAD, robotics, simulation, inspection, manufacturing, and interactive world authoring.

Recent work explores new opportunities at the intersection of vision-language-action models (VLAs) and geometric foundation models (GFMs) for 3D reconstruction, such as VGGT. While the resulting geometric VLAs often show improved performance, it remains unclear (i) if modern VLAs already have sufficient geometric understanding to start with, (ii) what is the best architecture to inject geometric understanding into a VLA, and (iii) what is the effect of other design choices that affect geometric VLAs. In this paper we provide a rigorous experimental analysis to shed light on these questions, for a specific choice of VLA (GR00T-N1.5) and GFM (VGGT). Our first contribution is to formalize prior work's intuition that current VLAs lack geometric understanding, by providing a rigorous analysis based on linear probing. The analysis quantifies, for the first time, the "geometric gap" between VLAs and GFMs. Our second contribution is to identify and compare different strategies to bridge GFMs with VLAs. We implement three different architectures, which differ in the way they inject geometry in the VLA, while keeping low-level implementation details as similar as possible, to ensure a fair comparison. Finally, we analyze the impact of non-architectural choices (e.g., training data, number of cameras, reconstruction quality) on the performance of the geometric VLAs.