Overview

Representation learning studies how to turn raw data—such as images, text, graphs, time series, and sensor signals—into features that make important patterns easier to use. Common approaches include contrastive and masked prediction, autoencoders, predictive latent models, sparse coding, and methods that impose structure such as geometry, hierarchy, or disentangled subspaces. Good representations can reduce label needs and support transfer across tasks, domains, and modalities, with evaluations spanning linear probes, retrieval, classification, and downstream control. A central tension is that compression and invariance can discard information a later task needs, so no single objective or representation geometry works best across settings.

Latest papers

855 papers91 papers in the last four weeks, down 14% on the four weeks before. 0.6% of all new papers.

Oct 7, 2026

Lossy Compressive Text Autoencoders

Our work explores learning a compressed latent representation of text, at the intersection of data compression and representation learning. We propose an autoencoder architecture that performs residual downscaling and upscaling of hidden representations along the time axis, with a residual low-dimension discrete bottleneck. We analyze our approach for different quantization methods, training objectives, and datasets. For different levels of compression, we evaluate the similarity between the original and reconstructed text both at the surface-level (BLEU) and at the semantic-level (LLM-based judge). Additionally, we evaluate our models on downstream question-answering and semantic text similarity benchmarks. Our approach results in compressed representations which are on par with lossless text compression algorithms at 2.24 bits per byte on web text data, while having good reconstruction and downstream task performance.
Oct 7, 2026

Scalable Patch-Level Self-Supervised Learning

Self-supervised learning (SSL) at scale produces powerful visual representations. However, most scalable SSL methods rely on ad hoc combinations of multiple objectives and stabilization mechanisms. Taking a step back, we ask if we can design a high-performing, yet principled SSL algorithm. Starting from the multi-view assumption, stipulating that task-relevant content is captured by the information common to different views, we construct an information-theoretic objective decomposing into interpretable terms. This derivation yields JEM, a student-teacher method that learns by aligning corresponding patch representations across views, explicitly regularized by information and structure preservation losses. JEM trains stably from 300M to 7B parameters, and, to our knowledge, is the first latent-space patch-level method demonstrated at 7B scale. Across all scales, JEM reaches strong performance on both global and dense probing tasks, on segmentation benchmarks consistently surpassing the DINOv2 algorithm, an influential foundation for today's strongest visual SSL methods. Notably, at 7B parameters, it exceeds the performance of DINOv3 on panoptic segmentation, despite being trained on 12×12\times less data without refinement stages. These results demonstrate that we can indeed design an SSL algorithm that learns strong representations, is principled and stable.
Oct 7, 2026

UniCSI Towards a Universal Wi-Fi CSI Encoder for Ubiquitous Human Sensing

Wi-Fi sensing promises to turn the everyday wireless signals that already surround us into ubiquitous sensors for human sensing. However, a fundamental obstacle is that CSI is acquired under diverse device-specific configurations, including different subcarrier counts, bandwidths, and carrier bands. Consequently, the resulting CSI tensors vary in both spectral resolution and tensor shape, making heterogeneous modeling challenging. Standard architectures struggle with such heterogeneity, forcing lossy pre-processing which compromises the underlying signal. To bridge this gap, we present UniCSI, a unified foundation architecture that directly operates on heterogeneous CSI while preserving the integrity of the native waveform. UniCSI hinges on two core innovations: (1) a physics-informed RF tokenizer that encodes each frequency channel based on its fractional position within the physical spectrum rather than rigid array indices. It preserves intrinsic spectral coherence and enables seamless, frequency resolution-agnostic processing across arbitrary sensing configurations. (2) a spectral aggregator that distills variable-length channel sequences into a fixed-size spectral signature, effectively decoupling the feature dimensionality from the physical subcarrier spacing. Extensive evaluations on a large-scale corpus of 25 heterogeneous public datasets, spanning 14 to 2048 subcarriers, 20 to 160 MHz bandwidth, and the 2.4 and 5 GHz bands, demonstrate that native heterogeneous ingestion substantially improves cross-domain transfer under both supervised and self-supervised training schemes, particularly in regimes where fixed-grid architectures fail to generalize.
Oct 6, 2026

Towards Financial World Modeling

Building a world model requires a state representation useful for planning and decision-making---potentially over tasks unknown at training time. In the context of financial markets, planning and decision-making may require a model to reason about market-wide conditions, asset-specific expected returns, liquidity, volatility, and cross-asset relationships. Yet financial representation learning has largely been evaluated on individual predictive tasks, oftentimes on a single time period using comparatively narrow datasets. We address this through three primary contributions. First, we introduce Market-1T, a dataset containing nearly one trillion observations across U.S. equities from 2008 to 2025 at 1 Hz resolution. Second, we develop and implement a rigorous evaluation protocol. Third, we conduct a systematic large-scale study of financial representation learning, comparing 18 encoder-training strategies across nearly two decades of market regimes. We evaluate learned representations both by their predictive utility on common finance tasks and through probes of latent structure. We find that encoders with similar predictive performance can organize market state very differently. Collectively, we establish a foundation for training and evaluating financial market representations in support of world models such as DINO-WM, V-JEPA 2, and LeWM.
Oct 6, 2026

PAIR: Bridging Perception and Action in Vision-Language-Action Models

Vision-language-action (VLA) models map visual observations and language instructions to continuous robot actions. This task requires a transition from representations that describe the scene and instruction to representations that support action generation. Many continuous-action VLAs leave this transition implicit and supervise it mainly through the final action-prediction loss. We introduce PAIR, a framework that learns a shared perception-action representation between these two spaces. During training, a Masked Action Autoencoder encodes expert action chunks into horizon-aligned Action Latent Tokens. A Bridge Module extracts task-relevant features from the current visual-language representations. PAIR aligns these features with the Action Latent Tokens to form Bridge Tokens that preserve task information and capture the structure of expert actions. The Bridge Tokens are then projected into the action-token space and injected into the initial Action Tokens, providing an action-ready starting point for Action Expert refinement. At inference, the autoencoder is removed, and the Bridge Tokens are generated only from the current observation and instruction. Experiments on LIBERO, LIBERO-Plus, and CALVIN ABC-D show gains for the evaluated OpenVLA-OFT and VLA-Adapter models. On LIBERO-Plus, PAIR raises VLA-Adapter's success rate from 59.1% to 64.2%. On CALVIN, it increases VLA-Adapter's average completed sequence length from 4.42 to 4.53. Across seven real-world tasks, PAIR raises OpenVLA-OFT's success rate from 51.4% to 65.0%. Representation analyses show that Bridge Tokens retain task information while making continuous-action information accessible before Action Expert refinement. These results support a shared intermediate representation as a useful interface between perception and action in continuous-action VLAs.
Oct 6, 2026

Symmetry-Aware Feature Learning: A Polynomial Separation for Multi-Index Models

We establish a polynomial sample complexity separation between symmetry-aware and symmetry-agnostic feature learning. We study growing-rank multi-index models with high-dimensional Gaussian covariates in Rd\mathbb{R}^d and r=Θ(dδ)r=Θ(d^δ) teacher directions forming a cyclic symmetry orbit, where 0<δ<1/20<δ<1/2. We compare three ways of exploiting this structure: architectural weight sharing, data augmentation over the full symmetry group, and learning without access to the symmetry. In particular, we analyze a symmetry-tied convolutional network, an untied network, and the same untied network trained with full-group data augmentation, using spherical online SGD with correlation loss. For a class of polynomial links with information exponent p≥3p\ge3, we prove matching sample complexity bounds up to logarithmic factors: the tied and augmented learners achieve weak directional recovery in Θ~(dp−1)\widetildeΘ(d^{p-1}) samples, whereas the symmetry-agnostic learner requires Θ~(rdp−1)\widetildeΘ(rd^{p-1}). For the pure quadratic Hermite link, the same separation holds for weak recovery of the teacher subspace, with sample complexities Θ~(d)\widetildeΘ(d) and Θ~(rd)\widetildeΘ(rd), respectively. Thus, full-group data augmentation matches the sample efficiency of architectural weight sharing, and both provide a polynomial advantage over training without symmetry. For p≥3p\ge3, the proof reveals a two-stage mechanism: fluctuations at initialization select one direction in the teacher orbit, after which localized growth amplifies its overlap to the weak recovery scale while competing overlaps remain near their initialization scale.
Oct 6, 2026

Which alloy composition,what process parameters? Inferring the recipe from optimized metallic microstructure and texture

The mechanical properties of a metallic alloy are set by its microstructure and texture: the size and shape of its grains and the orientation of their crystals. That structure is in turn set by a recipe, the alloy composition together with the processing parameters. Alloy development runs this chain forwards, tuning the structure until a target property is met. Running it backwards, from an optimized structure to the recipe that would produce it, still relies on expert knowledge. We ask whether this backwards step can be learned. On an in-house dataset of 107 magnesium alloy extrusion conditions across 14 alloys, each with optical micrographs and an X-ray texture measurement, we compare three descriptors of microstructure and texture: conventional grain and texture statistics, a vision embedding from a pretrained image encoder, and a graph neural network on the grain network. Each is paired with prediction heads for two tasks: the alloy composition given the process (Task A), and the process parameters given the composition (Task B). Under 5-fold cross-validation, the conventional descriptors identify the correct alloy for 65% of held-out conditions, against 17% for always guessing the most common alloy, while the learned embeddings stay below 30%. The process parameters are recoverable but noisier: compared with using the composition alone, the microstructure roughly halves the temperature error. Because only a few alloys were cast and only a few press settings were used, both answers are discrete, and heads that pick from these known options, while respecting their order, worked better than heads that predict a free value.
Oct 6, 2026

Preserving Unstable Modes Through Inverse Dynamics in JEPA World Models

Robotic systems often exhibit unstable modes, along which small perturbations and disturbances can cause unbounded growth unless corrected through feedback. Controlling such systems from high-dimensional visual observations requires representations that preserve these modes. Joint-embedding predictive architectures (JEPAs) provide a natural framework for learning such representations and their dynamics from visual data. However, we demonstrate that next step prediction combined with anti-collapse regularization does not guarantee that controllable unstable modes are preserved: the training loss can be minimized while these modes are collapsed, making stabilization from the learned representation impossible. To address this, we augment world-model training with an action reconstruction objective (i.e., an inverse dynamics loss) that encourages control-aware representations, namely, visual representations that preserve crucial features for control. We prove that exact action reconstruction makes the encoder injective on the finite-horizon reachable subspace. Thus, the encoder cannot discard any state direction reachable by an action sequence within HH steps. Moreover, we show that, as HH grows, the dominant eigenspace of the finite-horizon controllability Gramian converges to the controllable unstable subspace. We establish our theoretical results for linear systems and demonstrate empirically that our findings extend to nonlinear visual control tasks (CartPole, Walker2D, and PointMaze), highlighting the benefits of control-aware representation learning.
Oct 5, 2026

Structured Representation Learning for Behavior Cloning: How can we learn to safely control a nuclear power plant?

Learned models for industrial control are usually judged by aggregate accuracy, but accuracy at the component level does not guarantee safety once it is embedded in the system it is meant to serve. We study this gap on a behavior-cloning task: imitating an expert Nonlinear Model Predictive Control (NMPC) policy for load-following of a Pressurized Water Reactor (PWR), an industrial system with tight safety constraints. We propose a structured architecture encoding variables from each timescale into separate latent spaces, reflecting the physical decomposition of the system, before training a controller to imitate the expert on the product latent space. On long-horizon rollouts, separated embeddings improve both accuracy and feasibility compared with a shared-embedding baseline. Sensitivity analysis further shows that our model yields interpretable representations aligned with the system's physics. However, standalone deployment still leaves several percent of trajectories infeasible regardless of the architecture. Using our method to warmstart the NMPC optimizer rather than acting standalone, we recover full feasibility and near-optimal cost while still cutting computation time by ∼\sim15% relative to the expert controller, and even more for abrupt operating changes.
Oct 5, 2026

I-BFM: Reward-Conditioned Robust Humanoid Interaction via Unsupervised Reinforcement Learning

Behavioral foundation models (BFMs) have recently shown that a single humanoid policy can support diverse whole-body control, but extending such generality to physical interaction remains challenging. We introduce I-BFM, to our knowledge the first BFM for humanoid-object interaction. Rather than relying on task-specific policies or reference tracking, I-BFM learns a shared representation of the coupled dynamics among the humanoid, objects, and their contacts using forward-backward representations and unsupervised reinforcement learning. Given a downstream task reward, the same policy can be directly conditioned on a latent command to execute closed-loop interaction without task-specific policy optimization. To improve interaction control over different time scales, we further train the policy with both short-horizon interaction targets and longer-horizon goal targets. A single I-BFM policy performs carrying, pushing, and kicking, while also supporting goal reaching, motion tracking, stylistic control, and long-horizon task chaining. More importantly, it remains effective after large deviations from nominal execution: on Carry, I-BFM achieves 94.3% nominal success and retains 89.3% success after robot falls, compared with 1.3% for a planning-based baseline. Real-world experiments on a Unitree G1 further demonstrate diverse loco-manipulation behaviors, rapid recovery from interaction failures and external disturbances, and task chaining without task-specific retraining.
Oct 5, 2026

MagServo: Uncertainty-Resilient Hierarchical Magnetic Servoing via Learned Latent Representations

Magnetic navigation provides contact-free and line-of-sight-independent feedback for robotic systems, yet existing approaches typically rely on explicit pose estimation or direct use of raw magnetic measurements, making accurate control susceptible to modeling errors, measurement noise, and disturbances. This work presents MagServo, a hierarchical learning-based framework for robust 6-DoF magnetic servoing directly using the learned latent magnetic feature. MagServo learns uncertainty-resilient magnetic representations through masked reconstruction and captures state-dependent interaction dynamics between robot motion and latent magnetic transitions without analytical magnetic models or explicit Jacobian supervision. Based on the learned dynamics, a hierarchical controller combines nonlinear model predictive control for coarse approach with local Jacobian inversion for precise fine regulation. Extensive physical experiments demonstrate submillimeter and subdegree accuracy, achieving mean terminal errors of 0.386 mm and 0.479 degree for 6-DoF pose reaching. MagServo further outperforms a localization-based control baseline in complex trajectory tracking and maintains robust performance under unseen magnetic-source configurations without retraining. A supplementary video of the real-robot experiments is available at https://youtu.be/rZt1NUP1Mr0.
Oct 5, 2026

Character Identity is not Speaker Identity: KyaraBench and KyaraEmbed for Character Verification

A dubbed character keeps its identity while the voice actor changes, so character identity and speaker identity are distinct properties of one recording, yet speaker verification measures only the latter. To address this gap, we propose KyaraBench, a benchmark that scores character voice directly instead of speaker voice, built from 85 human-audited identities in a dubbed anime corpus. It poses two challenging conditions: one that swaps the performer under a fixed character, and one that fixes the performer under changing characters. Listening studies with 78 participants provide human reference scores for cross-performer verification and same-actor discrimination. Speaker-verification baselines show increased errors under these character-specific conditions. We then train KyaraEmbed, a compact encoder using multilingual character supervision, same-actor negatives, and a language-alignment term. The model achieves the best performance on all character-specific conditions in the main comparison. We release the benchmark, protocol, and encoder publicly.
Oct 5, 2026

From Pixels, Without Pre-training: Joint Generative and Self-Supervised Representation Learning in One Model

Strong image generation models are conditioned on class labels, aligned to frozen pretrained encoders, or built on separately trained autoencoders. While effective, generation then depends on supervision or pretraining: labels must be annotated, and encoders or autoencoders pretrained for the target domain. We study joint generative and self-supervised representation learning in a single model, enabling self-conditioned generation without labels or pretrained models. This is challenging because the objectives are mismatched: contrastive learning consumes clean augmented views and favors coarse, invariant semantics, while flow matching consumes noisy images and must preserve the fine detail and spatial layout that contrastive learning discards. We propose SCION (Self-conditioned Generation on Self-supervised representation), whose core is a single pixel-space encoder conditioned on the flow timestep and an embedding. For representation learning, this conditioning embedding is a learned global vector shared across images, with the encoder's [CLS] token yielding the semantic representation trained by the contrastive loss. For generative training, the conditioning embedding is the image's own [CLS] representation, while patch tokens pass through a decoder to predict the image. To sample without a reference image at inference, we jointly learn a prior over the embedding. Gradient-norm balancing and stop-gradient mechanisms enable joint optimization in one run. SCION is self-supervised and self-contained, with no labels or pretrained models. On ImageNet 256x256, with the JiT-B recipe and no representation guidance, SCION reaches 8.92 FID, surpassing class-unconditional iREPA, which aligns to pretrained DINOv2 (46.44), and RCG, which conditions on it (14.27). With JiT-L, SCION achieves 5.89 FID without guidance and 3.47 with representation guidance, outperforming RCG with the ADM recipe (6.24).
Oct 4, 2026

WiSPER: Pose-Supervised Predictive and Residual Flow Refinement For Multi-Person 3D Pose Estimation With WiFi CSI

Multi-person 3D pose estimation with WiFi channel state information (CSI) is challenging because reflections from different people overlap without directly identifying individual joints. Existing masked embedding objectives capture wireless relationships without explicit pose supervision, while structured decoders can retain coordinate errors. We propose WiSPER, a two-stage framework combining pose-aware predictive pretraining with conditional residual flow refinement. Pose-Aware Masked Embedding Learning (PAMEL) couples masked latent prediction with auxiliary pose-set supervision on the same CSI context, guiding the encoder toward joint localization from partial observations. Residual Flow refinement with Transformer (ReFT) generates a set of pose candidates to accommodate a variable number of people and refines each candidate through a conditional flow guided by its coarse coordinates and per-joint decoder features. Both stages use paired CSI and pose annotations during training, while inference requires only CSI. Experiments on the PiW3D dataset show that WiSPER achieves an overall mean per-joint position error of 63.72 mm, a 40.0% reduction relative to WiFi-JEPA. For experiments with two and three people, WiSPER reduces MPJPE by 42.1% and 38.1%, respectively. Pose-supervised pretraining configurations obtain lower errors than CSI-only JEPA, and enabling the trained residual refiner reduces overall MPJPE by 13.8-15.6% across the evaluated configurations.
Oct 4, 2026

Do Neural Networks Learn Structure-Preserving Maps? A Case Study in Latent-to-Hilbert Embeddings

We ask whether a neural network can learn a structure-preserving map from a compressed latent space to a Hilbert-space representation. Using an 8-dimensional autoencoder bottleneck on MNIST and nn-qubit product-state targets from PCA-based angle encoding, we report four findings. Although the target angles are generated by a nonlinear sigmoid transformation of the latent projections, the resulting mapping is well approximated by a linear function over the observed latent distribution: linear regression from zz to the true target angles achieves R2=0.98R^{2} = 0.98, while regression to the MLP's recovered angles achieves R2=0.91R^{2} = 0.91. The learned map's primary direction is strongly aligned with the target-induced direction, with cosine similarity 0.9890.989, while remaining nearly orthogonal to the input's principal direction, with cosine similarity 0.0020.002. The map is genuinely rank-4: removing any singular direction degrades inner-product preservation by 2.52.5--5.9×5.9\times despite a singular-value spectrum with two dominant and two small values. The learned subspace does not coincide with the PCA basis used to construct the target, and different random seeds recover the same primary direction but diverge in higher ranks. Finally, kernel ridge regression with an RBF kernel outperforms a tuned MLP (IP error 0.01440.0144 vs.\ 0.01970.0197), suggesting that for approximately linear structure-preserving mappings, classical kernel methods may be a simpler and more effective alternative.
Oct 4, 2026

EMG-GPT: Predictive Pretraining on Residual-Quantized EMG Tokens for Hand Pose Estimation

Surface electromyography (sEMG) is a low-power, cost-effective biosignal for hand-pose estimation and gesture classification. In this work, we examine whether self-supervised pretraining on sEMG can yield transferable representations for continuous hand-pose estimation. We introduce EMG-GPT, a causal transformer-based model that operates on discrete sEMG representations from a frozen residual vector quantization (RVQ) tokenizer and learns temporal dynamics through depth-autoregressive future-code prediction. The model combines within-frame integration with causal temporal modeling while preserving the geometry of the pretrained codebook. EMG-GPT shows competitive results in both Regression and Tracking tasks, supporting EMG-only pretraining as a viable approach for learning transferable sEMG representations.
Oct 3, 2026

Learning Discriminative Geometry for Drifting Models

Recently proposed Drifting Models shift iterative distribution refinement from inference to training, enabling effective one-step generation. However, their performance on complex image datasets depends strongly on the representation used to construct the drifting field: pixel-space drifting performs poorly, whereas pretrained feature spaces substantially improve sample quality for reasons that remain unclear. We trace this gap to the discriminative geometry of the representation, which determines sample weighting in kernel density estimation (KDE) and, consequently drift. We introduce persistent representation learning, which continuously learns a more discriminative representation geometry as the generator evolves across batches. We further establish a current-step gradient equivalence between the KDE ratio loss and drift regression loss under matched conditions, connecting density-ratio-based generator optimization to empirical drifting and motivating direct control of the drifting velocity. Across multiple datasets, our method learns effective discriminative representations directly from pixels and reduces FID by approximately 82−95%82-95\% over the original pixel-space Drifting Models, without pretrained encoders. Adapting pretrained representations and applying velocity clipping provide further gains.
Oct 3, 2026

Scaling of Wireless Foundation Models via Representation Diversity and Multi-Branch Architectures

Wireless foundation models learn representations from unlabeled radio signals for reuse across downstream tasks. Scaling model capacity is a common strategy for learning richer representations and improving downstream performance. However, its gains are less consistent in wireless self-supervised learning when pretraining data are limited. We investigate objective diversity as an alternative scaling axis: different self-supervised objectives emphasize different signal properties, and combining their representations can preserve more information to enable diverse tasks. We develop a fusion framework that combines frozen representations from independent encoders trained through reconstructive, predictive, and contrastive learning. This provides a reference for the benefits of diversity, but requires the maintenance of multiple encoders. To retain these benefits within the parameter budget of a standard single-objective encoder, we introduce a jointly trained multi-branch architecture with a shared trunk and objective-specific branches. We pretrain on a heterogeneous corpus of spectrogram and channel state information data and evaluate six downstream tasks that span communication, sensing, and positioning. At an equal output embedding dimension, fusion outperforms the evaluated single-objective encoders on all six tasks while using roughly one-third of their parameters. The multi-branch architecture retains much of the fusion benefit within a single-encoder parameter budget. Representation analyses indicate complementary contributions across objectives, with much of the added benefit retained in components orthogonal to the reconstruction representation subspace. These findings support objective diversity as an effective strategy for scaling wireless foundation models.
Oct 2, 2026

Author Representation Strategies for Zero-Shot Authorship Attribution: A Comparative Study of LLM-Based and Embedding-Based Approaches

Authorship Attribution (AA) requires capturing fine-grained stylistic characteristics, making it particularly challenging in zero-shot (ZS) settings where no task-specific supervision is available. In this work, we investigate the effect of author representations on ZS AA by evaluating a label-only prompting baseline together with three author representation strategies: representative writing samples, LLM-generated descriptions, and style embeddings (LISA). The first three approaches perform attribution using LLM prompting, while the embedding-based approach uses style embeddings with cosine similarity. We investigate the influence of prompt design and propose a two-stage embedding-based attribution framework that combines candidate space reduction with embedding-dimension selection. The results show that label-only ZS AA is ineffective, while incorporating author-specific representations consistently improves attribution performance. Among the evaluated approaches, the proposed two-stage LISA framework achieves the strongest overall performance, whereas LLM-generated style descriptions provide a substantially more compact representation of author style at the cost of some attribution performance. These findings demonstrate the importance of author representation in ZS AA, while indicating that current open-source LLMs remain insufficient for robust attribution without more effective representation learning.
Oct 2, 2026

Electronic Density versus Geometry for Machine-Learned Molecular Absorption Spectra

Molecular optical absorption spectroscopy provides a direct probe of electronic structure and is widely used for molecular identification, interpretation of photophysical behaviour, and planning of spectroscopy experiments. Calculating the absorption spectra using first-principle excited-state methods, however, is computationally demanding, at least compared to ground-state calculations, which limits their routine application across large molecular sets. Machine-learning (ML) surrogates can reduce this cost and allow rapid spectral prediction. However, their performance depends strongly on how molecular information is represented. Here, we compare using the ground-state electron density versus the molecular geometry as inputs to a ML model for predicting absorption spectra, for a training set of 6874 molecules selected from the QM7 dataset. For each of these molecules, the density was calculated using density functional theory (DFT) and the absorption spectrum was calculated using linear-response (LR) time-dependent DFT (TDDFT). Utilizing the ground-state density as the input to the ML model is motivated by the Hohenberg-Kohn and Runge-Gross theorems, and the fact that the ground-state density encodes information about bonding, charge localisation, and electronic delocalisation. Hence, it may be a more judicious starting point for the ML model compared to the geometry, as it effectively decouples the chemistry of the ground-state. The question we test is whether the benefits of using the density outweigh the (notprohibitive) penalty of requiring an additional single-point DFT calculation for the density. We find that the density-based convolutional neural network achieves a validation correlation of 0.9926, compared with 0.9795 for the best geometry-based graph model, reducing the residual decorrelation, by approximately 64%.
Oct 2, 2026

Muon Learns Facts Better: Understanding the Role of Spectral Orthogonalization

The Muon optimizer applies spectral orthogonalization to matrix-valued updates and has shown strong performance in large-scale neural network training, yet the mechanisms of this transformation in feature learning remain poorly understood. In this work, we investigate this question through a tractable factual-recall model, where a fact maps each subject-relation pair to an answer, and a linear transformer learns the subject- and relation-dependent information required to recover this mapping. The transformer is optimized with gradient flow (GF), spectral GF, or Sign GF, which are continuous-time limits of gradient descent, Muon, and Adam, respectively. Prior studies (Nichani et al., 2025) have shown that when the number of subjects exceeds the number of relations, GF learns relation-dependent information before subject-dependent information, producing a feature-separation phase during training. We characterize this separation with the learning times when the subject- and relation-dependent components of the prediction reach a target accuracy. With SS subjects and RR relations, GF has a learning-time ratio of Θ~(S/R)\widetildeΘ(\sqrt{S/R}), whereas Spectral GF reduces this ratio to Θ~(1)\widetildeΘ(1). In addition, for fixed SS and RR, the subject- and relation-dependent errors decay as 1/(Tlog⁡T)1/(T\log T) in training time TT under GF, but as exp⁡(−poly(T))\exp(-\mathrm{poly}(T)) under spectral GF. Finally, we show that GF and spectral GF are equivariant under orthogonal transformations of the token embeddings, whereas Sign GF is not: Different orthonormal embeddings can potentially produce no feature separation, a large feature-separation phase, or even a reversed learning order. These results provide a mechanistic view of how spectral orthogonalization can fundamentally reshape feature-learning dynamics.
Oct 2, 2026

AIGS: Adaptive Incremental Gating System for Online Representation Learning in Non-Stationary Data Streams

Real-time data streams in Web of Things (WoT) and edge computing environments often evolve through latent regime changes. For online representation learning under strict computational constraints, the central problem is resolving the stability-plasticity dilemma: keeping useful historical knowledge while rapidly reacting to concept drift. Existing methods employ fixed update schedules or rolling windows. However, they suffer from parameter ossification during sudden shifts and waste computational resources when the stream remains stable. This paper proposes the Adaptive Incremental Gating System (AIGS), a lightweight closed-loop state-aware adaptation framework. AIGS introduces the Shock Ratio, an endogenous residual feedback mechanism that normalizes current reconstruction error against recent variation. This signal drives a Continuous Plasticity Controller that smoothly interpolates between learning plasticity and memory retention. By treating representation learning as a closed-loop control mechanism, AIGS avoids catastrophic forgetting and maintains a strictly linear O(k⋅d)\mathcal{O}\left(k\cdot d\right) per-step complexity suitable for latency-sensitive edge devices. Experiments on real-world smart city dynamic streams-spanning traffic networks, meteorological systems, and industrial infrastructure-demonstrate distinct domain-dependent advantages. On Electricity Transformer Temperature datasets, AIGS achieves preventative early-warning lead times of 8.31 (ETTm1) and 9.88 (ETTm2) steps under gradual degradation. On Performance Measurement System traffic datasets, it shows significantly faster post-shift recovery after abrupt mutations. On the highly noisy Weather dataset, it improves anomaly recall while resisting stochastic noise overfitting. These findings establish AIGS as a practical, plug-and-play adapter for resource-constrained edge monitoring systems.
Oct 1, 2026

From Isolated Feature to Orbits: Discovering Music Concepts via Multi-SAE Alignment

How can we understand what a music foundation model has learned internally? Most interpretability approaches, such as probing and Sparse Autoencoders (SAEs), focus on identifying individual features with minimal structural assumptions. We argue that many concepts are better understood as structured relations rather than isolated features. This is especially prominent in music, where tonal structures are organized in the space of pitch and time. For example, concepts such as chords or keys are naturally expressed as structured sets (e.g., the 12 transpositions of a chord or the diatonic system within a key), rather than isolated features. In this study, we shift from feature identification to structure-based analysis, asking whether the learned inner representations of music foundation model emerge as organized structures over features. To this end, we introduce a framework that uses pitch transposition as an inductive bias to induce ordered orbits via multi-view SAE alignment. Concretely, we generate pitch-shifted input pairs and align their SAE representations to discover structured groups of pitch-related features. Experimental results show that this approach recovers orbit structures corresponding to chords, keys, and melodic patterns across two state-of-the-art music foundation models, while requiring only minimal grounding (e.g., a few anchor examples) to interpret entire concept families.
Oct 1, 2026

In-context Learning of Single-index Targets: Comparing Kernel and Feature Learners

In-context learning (ICL) enables a pretrained model to infer a task from demonstrations without updating its parameters. While much of the existing theory focuses on linear target functions, in this paper we study nonlinear cases by comparing two one-layer attention architectures on the same family of single-index tasks. A kernel learner first maps inputs through a fixed nonlinear feature map and then applies linear attention, whereas a feature learner applies attention to the original input, followed by a learned nonlinear readout. We derive predictions for their memorization and generalization errors using the replica method, retaining the effects of pretraining size, task-pool diversity, and training and inference context lengths. The resulting predictions closely match numerical experiments across a broad range of regimes. Our analysis yields phase diagrams that characterize when each architecture is advantageous as the amount of pretraining data, task diversity, and context lengths vary. We further identify qualitatively different context-length scalings for the two learners. Together, these results clarify how architectural choices interact with the dataset and govern nonlinear in-context learning.
Oct 1, 2026

Learning Commute-Time-Preserving World Models for Planning

World models allow agents to plan in latent space by choosing a sequence of actions that most reduces the distance to a given goal state. Thus, planning can benefit from latent representations whose distances mirror commute-times in the environment. The spectral embedding space of the graph Laplacian provides such a representation, if it obeys a specific eigenvalue-dependent scaling. Unfortunately, instantiating the graph Laplacian is intractable in large, continuous environments. Self-supervised learning offers a natural route to such commute-time-preserving embeddings at scale. However, here we show that existing methods, which commonly encourage isotropic representations to prevent representational collapse, tend to degrade the "correct" eigenvalue-dependent scaling, leading to an inaccurate representation of commute times. To address this problem, we introduce Commute-Time-Preserving World Models (CTWMs), combining a latent displacement predictor and a log-determinant regularizer that prevents collapse, which provably recover the correctly scaled Laplacian representation under reversible deterministic dynamics and at the predictor's fixed point. In numerical simulations, CTWM matches or outperforms LeWM, a task-agnostic baseline, on several complex, continuous goal-reaching benchmarks, while using half the parameters.
Oct 1, 2026

Not All Is Lost: Repairing Lossy User Preference States of Personalization Encoders

Personalization encoders compress evolving interaction histories into preference states used to rank items or condition text generation. A task head operating only on this state can miss useful evidence that remains in the frozen encoder's cached representations for individual timesteps. We study this recoverability gap and propose REPAIR, which compares cached representations with the current preference state in a compact learned coordinate space. It resolves corrective evidence over extended history, recent interactions, and localized bursts. It then selects which patterns at which timesteps contribute and adds their aggregate correction to the state before the task head. Encoder-host repair reuses representations from the existing forward computation without re-encoding the history. Across MovieLens, PENS, MIND, and Amazon Reviews 2023, training only REPAIR improves MRR and nDCG@10 for all twelve representative recommendation hosts while both encoder and task head remain frozen. Head-only finetuning of the same hosts yields smaller gains. For example, Mamba4Rec on MovieLens gains 3.96 MRR points, compared with 0.19 from head-only finetuning. Rank and temporal diagnostics support a compact, host-dependent corrective structure. In personalized generation, IMPerSumm improves the two reported weighted PerSEval variants, which assess responsiveness to user preference, by up to 25.23%. These results support post-compression state correction and distinguish the availability of preference evidence from its downstream use.
Sep 30, 2026

Reformulation-Contrastive Learning for Mixed Integer Programs

Mixed-integer linear programs (MILP) model many real-world decision problems, motivating machine-learning methods that exploit recurring structure to accelerate MILP solving. MILPs can admit many equivalent formulations: integrality-preserving changes of variables and the addition of redundant constraints can alter their formulations while preserving the optimization problem. We leverage these reformulations as a source of self-supervision for learning general-purpose representations of MILP variables and constraints. We characterize the affine reformulations that are valid for every input instance, and distinguish re-descriptions, which leave variables unchanged, from substitutions, which transform them predictably. Building on equivariant self-supervised learning, we introduce ReMILP (reformulation-contrastive MILP representation learning), which jointly trains a graph neural network and a hypernetwork to predict how variable embeddings transform under changes of variables. Without solver-derived labels, ReMILP learns representations that exhibit the intended invariance and equivariance on unseen problem classes. Across binary solution, constraint activity and integrality gap prediction, these representations carry task-relevant information when frozen and provide a useful initialization for fine-tuning.
Sep 30, 2026

CF-JEPA: Improving Robustness of JEPA World Models via Controllability Factorization

Controlling an agent with vision requires being able to separate useful information from irrelevant background information. JEPA-style latent world models seem like a natural approach for this, as they do not perform pixel-level reconstruction; however, they are still sensitive to these distractor signals and experience latent collapse. In this work, we introduce Controllability Factorized JEPA (CF-JEPA), a JEPA-style world model which splits the latent space into controllable and uncontrollable subspaces. This factorization allows us to capture all the distractor information into the uncontrollable region, while we use the control-relevant latent information for our task. With this, we show comparable performance across 2D and 3D control tasks under nominal conditions and improved performance under distracted conditions, where CF-JEPA is the only model that does not experience latent collapse. We also validate our model under distracted conditions for a simulated robot task, highlighting the practical application of such a scheme.
Sep 30, 2026

Rules Amortize, Pairings Don't: Linguistic Structure Determines What Latent Task Representations Can Replace In-Context Learning

In-context learning (ICL) can be amortized into latent objects (task vectors, function vectors, context vectors) that recover few-shot behavior at zero-shot inference cost, but recent theory shows a static vector acts as a single synthetic demonstration and must fail on high-rank mappings such as word-level bijections. We ask a linguistic version of this question: which linguistic operations can be amortized out of the prompt? We train a 2.6M-parameter network that reads the geometry of a few-shot support set (centroid, principal subspace, spectrum, computed once and cached) and produces an input-conditioned additive update to the query's residual stream at a mid-depth layer of a frozen GPT-2-large/XL. Across eight inflectional directions and one lexical relation, under a canonical split that bars inverted-pair leakage between directions, three regimes emerge. On forward inflection, where 10-shot ICL is strong (0.67-0.89) and extracted task vectors collapse (<=0.06), the transform matches ICL at strictly zero-shot per-query cost. On lemmatization directions, which frozen GPT-2 can execute but 10 demonstrations systematically fail to convey (ICL 0.13-0.48 at 1.5B), the transform is not capped by ICL at all: it reaches 0.78-0.92, up to +72 points over ICL (past to present: 0.85 vs. 0.13). On arbitrary pairings (antonymy) every amortizer plateaus near half of ICL at every scale, capacity, and seed tested. Controls show the support manifold acts as a causally necessary task fingerprint: wrong-task manifolds collapse accuracy to <=0.06, query-only variants cannot disambiguate tasks sharing an input space, and leave-one-task-out transfer is zero. Productive rules amortize into latent task representations, sometimes better than prompting can convey them; memorized pairings do not.
Sep 30, 2026

Gromov-Wasserstein Distillation for Inductive Multi-View Embedding

Gromov-Wasserstein multidimensional scaling (GW-MDS) learns low-dimensional representations from relational data but remains transductive, providing no explicit mapping for unseen samples. We introduce an inductive framework based on barycentric distillation. A GW-MDS teacher learns a latent support and an optimal transport plan from the training data, and barycentric projection converts the resulting coupling into sample-aligned targets. A neural student then learns an explicit out-of-sample mapping, avoiding additional relational-matrix construction and GW optimization at inference. We formulate the approach for single-view data and extend it to Mean-GWMDS and Multi-GWMDS teachers through consensus and selected-projection targets learned by a multi-view student with view-specific encoders. We also investigate a direct neural baseline trained solely with a GW objective. Experiments on synthetic and real-world data using Euclidean, geodesic, and cosine relations show that the distilled models preserve the teacher geometry on unseen samples and consistently outperform direct neural GW training in sample-indexed relational preservation. These results establish barycentric projection as an effective bridge between transductive GW embeddings and inductive neural mappings.