Knowledge Distillation

Ask about this topic

Overview

Knowledge distillation studies how a student model can learn from a teacher’s outputs, internal representations, examples, or behavior, often to reduce inference cost or transfer capabilities. Common approaches match softened class probabilities, hidden features, attention or relational structure, and use teacher-generated data; newer language-model methods also train on student-generated rollouts. The field spans model compression, cross-architecture and multimodal transfer, and settings such as federated learning, robotics, and low-resource tasks. A central challenge is choosing supervision that fits the student and remains reliable: stronger or denser teacher signals can help, but can also cause negative transfer, forgetting, or loss of capabilities not captured by the benchmark.

Latest papers

CardsList

830 papers164 papers in the last four weeks, up 95% on the four weeks before. 0.9% of all new papers.

Jul 8, 2026

Collaborative Synthetic Data Generation for Knowledge Transfer in Federated Learning

One-shot federated learning (OSFL) addresses the communication overhead of federated learning by limiting training to a single round, but doing so without sacrificing model quality is non-trivial, particularly when client data distributions diverge. Recent work has addressed this challenge by aggregating client knowledge on the server through the construction of transferable synthetic datasets or distillates. However, most of these methods lack formal privacy guarantees, leaving a gap in jointly achieving low communication, robustness to heterogeneity, and rigorous privacy. We propose FedKT-CSD (Federated Knowledge Transfer via Collaborative Synthetic Data), a framework inspired by neural image compression that closes this gap by leveraging publicly pretrained autoencoders as a shared latent space. Each client encodes its private data in a single forward pass, computes class-conditional latent statistics, and transmits these to the server. The server aggregates these statistics via secure aggregation, adds calibrated differential privacy noise, and decodes a synthetic dataset for training a global model and further downstream tasks. This design provides formal (ε,δ)(\varepsilon,δ)-differential privacy by construction, while keeping client-side computation and communication lightweight. Despite operating under privacy constraints, FedKT-CSD is competitive with and even outperforms non-private baselines across diverse datasets and heterogeneity settings, and scales to a large number of clients. Our code is available at: https://github.com/an7123/FedKT-CSD

Maximilian Andreas Hoefler, Karsten Mueller, Wojciech Samek

Fraunhofer Heinrich Hertz Institute · BIFOLD Technical University Berlin

Non-IID Federated LearningFederated LearningSynthetic Data GenerationFederated Learning AggregationDifferentially Private Federated LearningKnowledge Distillation+1
AskPDF
Jul 7, 2026

Enhancing deep learning models for time series classification via knowledge distillation

Deep learning has achieved remarkable success in various domains including time series analysis, computer vision and natural language processing. However, high computational and memory demands of state-of-the-art architectures pose challenges for deployment in resource-limited environments. Knowledge Distillation (KD) addresses this by transferring knowledge from a large teacher model to a smaller, more efficient student model while maintaining competitive performance. In this work, we investigate the effectiveness of KD for Time Series Classification (TSC) across three architectures: the classical Fully Convolutional Network (FCN), the convolutional Inception model and the transformer-based ConvTran model. We evaluate our approach on UCR Archive, the largest benchmark repository of time series datasets, by modifying architectural components such as convolutional filters, Inception modules and attention heads across the three architectures. Our results consistently show that KD most effectively benefits student models of intermediate complexity across all three architectures, with the distilled FCN student reducing parameters by a factor of 38, the distilled Inception student achieving nearly the same performance as the teacher with 42% fewer parameters and the distilled ConvTran student with 2 attention heads showing the most significant improvement through distillation. To encourage further research and reproducibility, we provide our implementation at https://github.com/MSD-IRIMAS/KD-4-TSC.

Javidan Abdullayev, Maxime Devanne, Jonathan Weber +1

IRIMAS, University of Haute Alsace, Frères Lumière, Mulhouse 68200, France · Faculty of IT, Monash University, Wellington Road, Melbourne 3800, Australia

Time Series ClassificationKnowledge DistillationNeural Network Compression
AskPDF
Jul 7, 2026

Few-Medoids: An Embarrassingly Simple Coreset Selection Method for Few-Shot Knowledge Distillation

Coreset selection aims to identify a small and highly representative subset of a massive dataset for efficient model training. The problem remains challenging even in the few-shot knowledge distillation (KD) setup, where a full-scale pre-trained teacher informs the student network. Typical sample selection strategies often struggle to surpass the random selection baseline. In this paper, we showcase few-medoids, an embarrassingly simple coreset selection strategy that chooses the samples closest to the centroid (average image) of each class. We present extensive KD experiments on four datasets, covering a wide range of image classification problems, and three teacher-student model pairs, comprising both convolutional and transformer networks. Although the proposed method is embarrassingly simple, our empirical results indicate that few-medoids is able to consistently surpass the random selection baseline, as well as the other coreset selection strategies. We therefore consider that few-medoids can be used as a drop-in replacement for commonly-used baselines (e.g. herding or k-center Greedy), in future research on coreset selection. To reproduce the reported results, we publicly release our code at https://github.com/CemilAndreiDilmac/Few-Shot-KD-Coreset.

Cemil-Andrei Dilmac, Florinel-Alin Croitoru, Radu Tudor Ionescu

Department of Computer Science, University of Bucharest, 15G Iuliu Maniu, Bucharest 061075, Romania

Training Data SelectionCoreset SelectionKnowledge Distillation
AskPDF
Jul 6, 2026

Patch Knowledge Transfer for Efficient AI-Generated Image Quality Assessment

With the rapid advancement of image generation technologies, perceptual quality assessment of AI-generated images has emerged as a crucial research direction in computer vision. The core challenge of this task lies in achieving efficient quality assessment for massive generated images. Current mainstream approaches exhibit two key limitations: 1) Methods employing complex feature extraction strategies, while improving performance, incur prohibitive computational costs that hinder real-time inference; 2) Simple image scaling-based solutions, despite their computational efficiency, demonstrate significantly inferior assessment accuracy. To address this critical issue, we propose Patch Knowledge Transfer (PKT), a knowledge distillation-based optimization framework that achieves synergistic optimization of visual representation capability and inference efficiency through an innovative multi-level knowledge transfer mechanism. Specifically, we design a dual-model architecture: a teacher model with local-global hybrid processing provides high-quality supervision signals, while a student model relying solely on global processing efficiently inherits the teacher's representation capacity through multi-level supervision. Extensive experiments conducted on 4 AIGIQA databases demonstrate that the PKT framework enables the student model to maintain performance comparable to the teacher while reducing computational costs by 67.7%. Furthermore, compared to existing methods, our approach achieves a superior balance between model efficiency and assessment accuracy.

Jiquan Yuan

School of Software and Microelectronics, Peking University, Beijing, China

Image Quality AssessmentImage Generation EvaluationKnowledge Distillation
AskPDF
Jul 6, 2026

Multi-Teacher Contrastive Distillation for Edge-Efficient Pathology Foundation Models

Computational pathology foundation models (PFMs) have advanced whole-slide image analysis. However, their size and inference cost hinder local deployment in pathology departments. We propose MuCoDi, a pretraining framework that distills frozen tile embeddings from multiple PFMs into compact edge-oriented encoders. Instead of regressing individual teacher features, MuCoDi trains lightweight MobileOne and RepViT students with a contrastive distillation objective adapted from MoCo v3, where cached Virchow2, UNI2, and H-Optimus-1 embeddings replace momentum-encoder keys. We pretrain students on 14.3M TCGA tiles from only 11.8K WSIs and evaluate frozen encoders on 23 clinically curated downstream classification tasks. RepViT-based MuCoEdge students retain near-teacher performance while reducing model size by orders of magnitude: MuCoEdge-R2.3 and MuCoEdge-R1.5 reach 71.0% external AUROC, within 0.8 percentage points of the best teacher (Virchow2, 71.8%), while MuCoEdge-R2.3 obtains the best external F1 and the second-best AUPRC (51.8% and 53.3%). MuCoEdge-R1.0 reaches 70.9% AUROC with only 6.4M parameters and 1.12 GFLOPs. On a Raspberry Pi 5, sub-million-parameter MobileOne students achieve up to 605-fold single-tile speedup over Virchow2 while retaining 66.5% to 66.9% external AUROC, demonstrating that PFM-quality pathology representations can be moved toward practical edge deployment. Code is available at https://anonymous.4open.science/r/mucodi-6243.

Tim Lenz, Maurice Heide, Marco Gustav +2

EKFZ for Digital Health, TU Dresden, Dresden, Germany · Pathology, University of Augsburg, Augsburg, Germany · Bavarian Cancer Research Center (BZKF), Augsburg, Germany +3

Computational PathologyPathology Foundation ModelsHistopathology Image ClassificationEdge MLKnowledge Distillation
AskPDF
Jul 6, 2026

Trust Region Policy Distillation

Big goals are hard to achieve all at once; breaking them into small steps is wiser. We present Trust Region Policy Distillation (TOP-D), which transforms the notoriously unstable, high-variance On-Policy Distillation (OPD) into a stable training paradigm by dynamically constructing a proximal teacher. Theoretically, we establish a rigorous framework demonstrating that TOP-D inherently controls gradient variance. By providing a formal global convergence analysis alongside a monotonic improvement bound, we mathematically formalize the reliability and stability of the overall training dynamics. Empirically, TOP-D dramatically enhances training stability, sample efficiency, and final performance on mathematical reasoning tasks. More importantly, TOP-D introduces zero additional computational overhead, positioning itself as a promising alternative to the well-established OPD paradigm.

Zhengpeng Xie, Li Lyna Zhang, Zeke Xie +1

On-Policy DistillationPolicy LearningPolicy DistillationKnowledge DistillationRL Stability
AskPDF
Jul 6, 2026

Displacement Preserving Relational Distillation for Robust Medical Segmentation

Accurate 3D medical segmentation is limited by anatomical variability and high computational costs. While knowledge distillation (KD) offers a route for model compression, conventional methods often fail to preserve complex structures and are overwhelmed by background noise. We propose Displacement-Preserving Relational Distillation (DPRD), which distills latent anatomical trajectories via vector based alignment to preserve the orientation and relative scale of the teacher's manifold, and prevents signal dilution by anchoring distillation in task-relevant structures. Integrated into nnU-Net, DPRD outperforms established baselines on ISLES 2022 and AMOS 2022 benchmarks. Notably, on the AMOS dataset, DPRD achieves a Dice score of 85.46%, edging out the high-capacity MedNeXt teacher while significantly reducing boundary errors. Despite utilizing only ~5% of the teacher's parameters and ~3% of its FLOPs, our approach maintains high structural consistency. This provides a robust, efficient solution for deploying high performance segmenters in resource-constrained clinical environments. Code: https://github.com/ClinicaAlpha/DPRD-3D-MedSeg

Zhicheng Ding, Xinyu Chu, Jung Im Choi +5

Bowling Green State University, Bowling Green, OH, USA · University of Alabama at Birmingham, Birmingham, AL, USA · The University of Findlay, Findlay, OH, USA +3

Model CompressionRelational Knowledge Distillation3D Medical Image SegmentationKnowledge Distillation
AskPDF
Jul 5, 2026

Covert Trait Propagation Is Representation Alignment: Mechanistic Evidence from Hidden-Channel Distillation

A student model trained on pure uniform noise can still inherit its teacher's digit-classification ability, provided the two share initialization. Previous work proves this transfer is guaranteed when the teacher's learning rate is small enough, but does not explain where in the network the channel lives or what sets its capacity. Working in an MLP distillation setting on MNIST, we show these channels are not purely informational: geometric alignment gates access to the information the channel carries. Shared initialization makes the output projection W_2 a common coordinate key, and KL gradients reshape the student's input projection W_0 until its hidden representations align with the teacher's. We call this covert trait propagation (CTP). Five experiments support this mechanism: channel closure tracks weight drift, not teacher accuracy; freezing W_0 destroys transfer while freezing W_2 leaves it intact; multi-teacher ensembles cancel out despite each teacher carrying comparable label information; and linear centered kernel alignment (CKA) tracks student accuracy at r=0.98 across a continuous initialization sweep. Applying the same geometric lens to cross-token behavioral entanglement (CTBE) in instruction-tuned LLMs, we find the effect appears to be activated by alignment training, acting on an inherited substrate, and that the standard log-ratio metric produces an apparent frequency bias that is largely a circularity artifact.

Kargi Chauhan, Aditya Shah

University of California, Santa Cruz · 2Google

Mechanistic InterpretabilityKnowledge DistillationRepresentation Alignment
AskPDF
Jul 5, 2026

SAD-LoRA: Spectral Alignment for Low-Rank Knowledge Distillation

Distilling a fine-tuned teacher into a LoRA-adapted student is a standard recipe for parameter-efficient compression, but output-level KD does not explicitly control which rank-rr weight subspace the adapter occupies. We propose SAD-LoRA (Spectral Alignment Distillation), which selects this subspace from the data-weighted student-space reference update \DWT\Sigx1/2\DWT\Sigx^{1/2} and maintains it during training via a differentiable principal-angle loss on \colspan(B)\colspan(B). We show that the data-weighted distillation error decomposes exactly into subspace misalignment, within-subspace coefficient mismatch, and irreducible rank residual; standard KD can affect the first term only indirectly through output gradients. On controlled synthetic problems with a flat teacher spectrum, SAD-LoRA reduces the subspace-misalignment term from 51%51\% to nearly zero and lifts final subspace alignment from 0.490.49 to 1.001.00. On RoBERTa-large to RoBERTa-base distillation across six GLUE tasks, SAD-LoRA improves rank efficiency: at r=4r{=}4, it matches or beats the strongest included spectral baseline on five of six tasks, and at r=8r{=}8 it gives the best result on SST-2 and CoLA. Ablations identify subspace alignment as the load-bearing component, while coefficient matching is auxiliary.

Omer Tariq, Syed Muhammad Raza, Jeongbae Son

Neubility Inc, Seoul, South Korea

Low-Rank AdaptationLow-Rank CompressionLanguage Model DistillationKnowledge Distillation
AskPDF
Jul 5, 2026

XS-VLA: Teaching Tiny Vision-Language-Action Models with Spatial Supervision and Demonstration Conditioning

How can richer training supervision improve robot control while keeping the deployed policy compact? We present XS-VLA, a staged training framework using teacher-derived spatial labels and demonstration-conditioned action learning. Coarse-Grained Spatial Distillation (CSD) initializes the backbone through an auxiliary region-label task. Latent Flow Matching (LFM) then conditions an action-space velocity field on a demonstration latent, using KL regularization while jointly optimizing the backbone and action modules. The deployed policy contains 243.99M parameters and operates without the teacher or posterior encoder. XS-VLA achieves 90.25% average LIBERO success in each of two training seeds, compared with 86.00% for a SmolVLA-256M base trained under our settings. Ablations examine both training stages through matched image pretraining and Huber/MSE controls. On three Mobile ALOHA tasks, average strict success increases from 21.7% to 65.0%. These results demonstrate the control utility of auxiliary representation initialization and regularized demonstration-conditioned flow learning for compact VLA~policies.

Iok Tong Lei, Ying Jie Yap, Wei Huang +5

Department of Computer Science and Technology, Tsinghua University, Beijing, China. · National College for Excellent Engineers, Beihang University, Beijing, China. · Independent Researcher +1

Robot Policy LearningLanguage-Conditioned Robot ManipulationEfficient VLA ModelsVision-Language-Action ModelsVision-Based Robot ControlKnowledge Distillation
AskPDF
Jul 4, 2026

ContiStain: Cross-Domain Relation-Preserving Distillation for Continual Multi-Domain Virtual IHC Staining

A unified multiplex virtual staining model enables scalable and non-destructive multiplex analysis from H&E slides while promoting parameter efficiency, shared pathological knowledge, and consistent cross-biomarker representations. However, in clinical practice, data for new biomarkers are typically acquired sequentially over time. Fine-tuning on such temporally arriving data leads to severe performance degradation on previously learned biomarkers, as sequential optimization disrupts the structured relationships among biomarker representations in the latent space. To address this issue, we propose ContiStain, an IHC multi-domain relational distillation framework for continual virtual staining. We first (i) construct a domain-aware structured feature space using a mixture-of-experts (MoE) feature extractor to reduce representation interference across biomarker domains. Based on this stabilized feature space, we then (ii) propose a relation-preserving distillation strategy that explicitly enforces the consistency of cross-domain token-level cosine similarity matrices between learned biomarker domains during continual adaptation. By maintaining cross-domain structural coherence, ContiStain mitigates forgetting while retaining adaptability to new domains. Experiments on the MIST dataset under a four-domain sequential virtual IHC staining setting show improved stability, reducing FID and ConchFID by 11.1 and 60.9 compared to sequential fine-tuning, enabling scalable and robust multi-domain virtual staining. Code is released at https://github.com/ccitachi/ContiStain.

Fuqiang Chen, Yifeng Wang, Hongpeng Wang +1

School of Computer Science and Technology, Harbin Institute of Technology (Shenzhen), Shenzhen, China · School of Computer Science and Technology, Tsinghua University, Beijing, China

Computational PathologyVirtual StainingDomain-Incremental LearningRelational Knowledge DistillationKnowledge Distillation
AskPDF
Jul 3, 2026

Teacher Supervision over Representation Equivalence Classes

Knowledge distillation is usually framed as a choice of what to match in the teacher - its logits, hidden features, or sample relations - which presupposes that the teacher's representation has absolute coordinates to match. It does not: a pretrained representation is identifiable only up to an orthogonal-and-isotropic-scaling equivalence class, so a student should learn the teacher's equivalence class, not its features. The organizing fact is that capability is the teacher's output function, a class invariant that factors through the quotient by the class action, so an objective recovers capability exactly when it is defined there. This makes absolute feature matching ill-posed, and admissible supervision a matter of targeting class invariants (Gram structure, CKA, principal subspaces) or aligning coordinates first, unifying feature matching, relational distillation, alignment, and grafting in one geometric account. We validate our framework on Qwen2.5 and Llama-3.1. A restoration study recovers a corrupted model's representation (CKA ~ 0.99) but not its capability, and an ablation isolates the cause: output-function (logit) matching drives capability, while matching hidden representations aligns geometry without restoring function. Recovery is confined to the corpus-covered region, and a graft study confirms that boundary overlap predicts transplant success but is necessary, not sufficient.

Sang Il Han

Korea University of Technology and Education · Biomonster Co.

Representation IdentifiabilityLanguage Model DistillationRepresentation Geometry in Language ModelsKnowledge DistillationRepresentation Alignment
AskPDF
Jul 2, 2026

ComplexMimic: Human-Scene Interaction Imitation in Complex 3D Environments

Physics-based Human-Scene Interaction (HSI) imitation learning is crucial for embodied intelligence as it bridges the gap between kinematic 3D motions and real-world dynamics. However, most existing methods focus on simplified scene settings, leaving complex environments largely unexplored, which limits their applicability in real-world scenarios. In this paper, we focus on HSI mimicry in complex environments. Under this complex setting, we observe an inherent trade-off between successfully performing interaction and maintaining natural, physically plausible motions. To address this challenge, we propose ComplexMimic, a framework that reconstructs diverse HSI by interpreting imperfect MoCap data. First, we introduce a Dual Flow Strategy, which learns two complementary experts: an imitation expert for accurate motion tracking and an interaction expert for collision-aware adaptation in complex scenes. Second, naive multi-expert distillation, which treats all experts equally, often under-samples challenging behaviors, limiting effective learning. To mitigate this issue, we propose a difficulty-aware distillation strategy that adaptively weights supervision and prioritizes hard-yet-learnable trajectories guided by failure statistics and learning progress signals. Extensive experiments on three benchmark datasets demonstrate that our approach outperforms current state-of-the-art methods.

Lu Pan, Hongwei Zhao

College of Computer Science and Technology, Jilin University, China · Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education, Jilin University, China

Human Behavior SimulationHuman Motion GenerationImitation LearningKnowledge Distillation
AskPDF
Jul 2, 2026

SFKD: Spatial--Frequency Joint-Aware Heterogeneous Knowledge Distillation via Multi-Level Wavelet Spectral Interaction

Most existing knowledge distillation methods focus on homogeneous models (e.g., CNN-to-CNN), thereby overlooking the flexibility and potential of knowledge transfer across heterogeneous models. Due to intrinsic inductive bias discrepancies between heterogeneous models that cause spatial distribution inconsistencies, prior heterogeneous distillation methods often weaken or discard spatial information in heterogeneous representations. However, the spatial information in representations often encodes transferable global structural semantics as well as architecture-specific local details, and therefore should not be directly ignored. To better leverage the spatial information encoded in heterogeneous representations, we propose a Spatial-Frequency Joint-Aware Heterogeneous Knowledge Distillation framework (SFKD). By leveraging the complementary properties of wavelet transform spatial locality and Fourier representations in characterizing global energy distributions, we first apply multi-level discrete wavelet transform to explicitly decouple spatial information. The resulting wavelet sub-bands are further refined by a dual-stream dual-stage refinement module, and finally combined with a Gaussian-filtered frequency loss to selectively capture informative global information. Extensive experiments on multiple benchmark datasets under both homogeneous and heterogeneous models demonstrate the superiority of our method.

Cuipeng Wang, Haipeng Wang

Key Laboratory for Information Science of Electromagnetic Waves, Ministry of Education, Fudan University, Shanghai, China · Discipline and Technology Center of Microwave Vision Intelligent Sensing, Fudan University, Shanghai, China

Cross-Architecture Knowledge DistillationKnowledge Distillation
AskPDF
Jul 2, 2026

Geometric Foundation Model Distillation for Efficient Lunar 3D Reconstruction

Large 3D foundation models such as MASt3R achieve state-of-the-art stereo reconstruction but are computationally demanding for deployment under strict hardware constraints -- a critical limitation in domains such as planetary exploration, where onboard computing is severely restricted. We study how far such models can be compressed through knowledge distillation, using lunar stereo reconstruction as a challenging and practically relevant case study. Starting from a 688M-parameter MASt3R teacher fine-tuned on lunar imagery, we distill its dense geometric predictions into a family of lightweight students spanning different encoder types (CNN vs ViT), decoder widths and depths, and training strategies. To bridge the dimensional mismatch between teacher and student, we propose a structured SVD-based initialization that projects the teacher's decoder weights into the student's smaller latent space, yielding a warm start that significantly improves convergence and final performance. Based on our results on lunar data, we can obtain a distilled student that retains most of teacher's reconstruction accuracy while reducing the model size up to 7 times, and even outperforms a baseline trained directly with sparse ground-truth annotations. Beyond compression, our study highlights both principles and practical insights for distilling geometric foundation models: a convolutional encoder underperforms transformer-based alternatives (though pretraining availability remains a confounding factor), preserving encoder capacity is more critical than maintaining a large decoder, feature-level distillation consistently outperforms output-only supervision, and SVD-based initialization improves optimisation stability. These findings provide practical guidelines for deploying 3D reconstruction models in resource-constrained environments.

Clémentine Grethen, Florient Chouteau, Géraldine Morin +1

1 1 IRIT, University of Toulouse, France · 2 2 Airbus Defence and Space, Toulouse, France

3D Foundation Models3D ReconstructionKnowledge Distillation
AskPDF
Jul 2, 2026

C2E: Boosting Ego-Only 3D Object Detection via Multi-Teacher Contrastive Knowledge Distillation

LiDAR-based 3D object detection is essential for autonomous driving systems. However, traditional Ego-only Perception (Eo-Perception) suffers from limited perspective and occlusions in a complex outdoor environment, leading to performance bottlenecks. Recently, research on multi-agent Collaborative Perception (Co-Perception) has demonstrated excellent performance, but high communication costs and accumulated pose error hinder its application. To address this, we explore a novel C2E (Co-Perception to Eo-Perception) paradigm through the Multi-to-Single (M2S) agent contrastive knowledge distillation framework. Our M2S framework first designs Multi-Level Feature Enhancement module to provide more stable features, and introduces Auxiliary Point Cloud Reconstruction and Multi-Teacher Contrastive Distillation mechanisms to mitigate domain gaps in point cloud and feature distributions within the C2E paradigm. Benefiting from this, our M2S can retain the excellent performance of collaborative perception while effectively avoiding the drawbacks, such as communication delays and positioning errors. Extensive experiments on the V2XSet, V2V4Real and DAIR-V2X datasets show the effectiveness and generalizability of our M2S framework when combined with the state-of-the-art CoSDH model and other excellent 3D detectors. Our M2S framework can deliver up to a 8.64% improvement in 3D mAP performance without introducing any communication costs.

Jinlong Wang, Xun Huang, Qiming Xia +2

Fujian Key Laboratory of Urban Intelligent Sensing and Computing, Xiamen University, China · Key Laboratory of Multimedia Trusted Perception and Efficient Computing, Ministry of Education of China, Xiamen University, China · Zhongguancun Academy, China

Lidar-Based 3D Object Detection3D Object DetectionCooperative PerceptionKnowledge Distillation
AskPDF
Jul 1, 2026

Token-level Response-visual Attention Guidance for Multimodal LLMs Knowledge Distillation

While knowledge distillation (KD) is widely adopted for training lightweight models by leveraging supervision from larger teacher models, relying solely on output token distributions has proven insufficient for compressing Multimodal Large Language Models (MLLMs). Since output tokens are a byproduct of the model attending to visual inputs, prior works have explored explicitly distilling attention to provide a direct supervisory signal. While promising, the precise utility of which attention signals to distill remains under-explored. In this work, we challenge the conventional reliance on prompt-to-vision attention by revealing that downstream performance correlates strongly with response-to-vision attention similarity to the teacher, but negligibly with that of prompt-conditioned attention. Furthermore, we observe that attention distributions exhibit significant variance across individual tokens, indicating that a uniform distillation objective is suboptimal. To this end, we introduce Token-level Response-visual Attention Guidance (TRAG), a distillation objective that 1) shifts the focus to response-to-vision signals and 2) employs token-specific objectives by adaptively weighting the Kullback-Leibler divergence based on attention entropy, effectively guiding the student to mirror the teacher's precise visual focus. Extensive experimental results on multiple benchmarks demonstrate that TRAG significantly outperforms prior distillation baselines.

Jaehyun Jang, Eunseop Yoon, Hee Suk Yoon +3

Korea Advanced Institute of Science and Technology, Daejeon, Republic of Korea · University of Illinois Urbana-Champaign, Champaign, USA

Cross-Modal Knowledge DistillationVLM DistillationMultimodal Large Language ModelsSelf-Supervised Knowledge DistillationKnowledge Distillation
AskPDF
Jul 1, 2026

OnPoint: Offline-to-Online Multi-Level Distillation for Point-Supervised Online Temporal Action Localization

Temporal Action Localization (TAL) typically relies on segment annotations or offline access to full videos, limiting scalability and online use. We introduce Point-Supervised Online TAL (POTAL), which localizes actions in streaming videos using only one temporal point per instance. To solve POTAL, we propose OnPoint, an offline-to-online multi-level distillation framework that transfers knowledge from a point-supervised offline teacher to an online student via (i) pseudo-segment instance distillation, (ii) class-activation sequence distillation, and (iii) anticipatory window-level distillation. We further improve robustness by incorporating the original point labels into student training and by refining anchor decoding with actionness-guided attention calibration. Experiments on five datasets show OnPoint consistently outperforms strong baselines, establishing a solid foundation for POTAL.

Sakib Reza, Gauri Jagatap, Mohsen Moghaddam +2

Northeastern University, Boston, MA 02115, USA · Georgia Institute of Technology, Atlanta, GA 30332, USA · Dolby Laboratories +1

Knowledge DistillationTemporal Video UnderstandingTemporal Action Localization
AskPDF
Jun 30, 2026

TallyTrain: Communication-Efficient Federated Distillation

Federated learning is bandwidth-bound on two orthogonal axes: model size, which limits how often parameter-averaging methods can afford to merge, and class count, which makes per-probe soft-label distillation prohibitive at large vocabularies. Both ceilings tighten as modern systems scale. We collapse the class-count axis to ⌈log⁡2C⌉\lceil \log_2 C \rceil bits per probe by transmitting only each peer's arg⁡max⁡\arg\max class index, where CC is the number of output classes. The resulting protocol, TallyTrain, is not merely compressed: under non-IID training it can be preferable to soft-label distillation, because under-trained peers are confidently wrong and majority voting filters this noise where soft-label averaging amplifies it. Across standard benchmarks, TallyTrain matches or beats soft-label distillation at up to three orders of magnitude less communication. We also relax the model-size axis: we compose the cheap hard-label consensus with sparse parameter merges to obtain a bandwidth-bridge variant, which Pareto-dominates every tested operating point of the standard FedAvg, FedProx and FedDF baselines.

Radhakrishna Achanta, Will Reed

Cisco Systems Inc.

Federated Learning AggregationCommunication-Efficient Distributed TrainingFederated Knowledge DistillationNon-IID Federated LearningKnowledge DistillationFederated Learning
AskPDF
Jun 30, 2026

Benchmarking Federated Learning and Knowledge Distillation for Point Cloud Classification

Deploying 3D point cloud analysis in privacy-sensitive, resource-constrained settings faces two barriers: data cannot be centralized, and models must run on limited edge hardware. We present a multi-seed benchmark jointly evaluating federated learning (FL) and knowledge distillation (KD) for 3D point cloud classification. It spans 13 FL algorithms and 10 KD objectives (a 130-pair cross-product) across 504 training runs, evaluated on ModelNet40 and a clinical craniosynostosis dataset. We report three findings. First, under extreme non-IID label skew, standalone FL degrades sharply: on ModelNet40, the strongest method reaches 76.32% against a 92.26% centralized reference; on clinical data, the best reaches 75.83% against 100%. Second, distillation successfully compresses the teacher into a student 74.51% smaller and roughly twice as fast at inference, often matching or surpassing the teacher. Third, the combined pipeline exposes an evaluation pitfall: when distillation keeps a hard-label cross-entropy term on a labeled proxy split, a collapsed federated teacher (8.50%) paired with Logit-MSE still yields a 92.94% student. This 84.4-point gap reflects the proxy labels rather than the federated model, reusing the very labels whose privacy motivated federation. Objectives without hard labels instead track teacher quality (r≈0.99r \approx 0.99) and collapse when the teacher does. We therefore recommend evaluating FL-KD pipelines with label-free distillation so reported accuracy reflects the federated teacher, not the proxy.

Aizierjiang Aiersilan

University of Macau

Data LeakagePoint Cloud ClassificationFederated Knowledge DistillationNon-IID Federated LearningKnowledge DistillationFederated Learning
AskPDF
Jun 30, 2026

Improving Certified Robustness via Adversarial Distillation

Certified training aims to produce models whose predictions can be formally verified against adversarial perturbations, typically by optimising upper bounds on the worst-case loss over an allowed perturbation set. For neural networks, certified training methods based purely on tight relaxation bounds produce networks that are amenable to certification, but sacrifice standard accuracy. Conversely, adversarial training often yields stronger empirical robustness and standard accuracy, but the resulting models are generally difficult to certify with neural network verifiers. Recently, the literature has shown that better standard-certified accuracy trade-offs can be achieved by combining adversarial training objectives with loose over-approximations based on Interval Bound Propagation (IBP), effectively interpolating between lower and upper bounds of the worst-case loss. Building on this, we introduce AD-CERT, a certified training objective that combines adversarial distillation with an IBP upper bound. We show that distilling adversarial information over the logit space from an empirically robust teacher provides an effective lower bound surrogate for certified training, with AD-CERT achieving state-of-the-art certified performance on several robustness benchmarks for ReLU networks. Furthermore, in a unified setup, distilling adversarial information at the logit-level is shown to improve certified accuracy over a robust feature-space distillation objective by up to 5.40 percentage points. Our code has been made available at https://github.com/matteo-melis/ad-cert.

Matteo Melis, Jesus Martinez Del Rincon, Vishal Sharma

School of EEECS Queen’s University Belfast

Neural Network RobustnessCertified RobustnessKnowledge DistillationAdversarial Knowledge Distillation
AskPDF
Jun 30, 2026

Holonic Active Distillation for Scalable Multi-Agent Learning in Multi-Sensor Systems

The rapid expansion of sensor-based networks introduces major challenges in scalability, adaptability, and knowledge transfer, especially in open environments where new subsystems can dynamically join or leave. In this work, we propose a Holonic Active Distillation architecture within a Holonic Multi-Agent System (HMAS) to address these issues. Our approach integrates Clustered Stream-Based Active Distillation (CSBAD), a framework in which specialized student models collect local data, query pseudo-labels from teacher models, and cluster into groups of similar sensors. Results show that the holonic organization balances local specialization with global generalization, while efficiently adapting to sensor departures and re-integrations. We also analyzed trade-offs among incremental model updates, system reorganization, and scalability limits. Our findings highlight the advantages of holonic learning for multi-sensor systems while identifying key challenges related to model drift and long-term adaptation.

Dani Manjah, Tim Bary, Benoît Macq +1

Institute of Information and Communication Technologies, Electronics and Applied Mathematics (ICTEAM), UCLouvain, 1348 Louvain-la-Neuve, Belgium · Université de Technologie de Belfort Montbéliard, UTBM, CIAD UR 7533, F-90010 Belfort cedex, France

Multi-Agent SystemsMulti-Teacher Knowledge DistillationKnowledge Distillation
AskPDF
Jun 30, 2026

Knowledge Distillation from Large Reasoning Models to Compact Student Models: A Case Study on the John O Bryan Mathematics Competition

This paper investigates knowledge distillation from a large reasoning model (DeepSeek-R1) to a compact student model (Qwen2.5-7B). Using historical problems from the John O'Bryan Mathematics Competition at Northern Kentucky University (2011-2025), we build a Chain-of-Thought (CoT) training corpus through a dual-agent framework. The dataset is used to fine-tune the student model with Low-Rank Adaptation (LoRA) on Apple Silicon hardware using the MLX framework. The base Qwen2.5-7B model achieves 64.67% accuracy on competition problems, while the DeepSeek-R1 teacher achieves 91.40%. An initial 1,000-iteration training run revealed severe overfitting, with validation loss reaching a minimum at iteration 200 before rising steadily. Based on this finding, we ran five independent training runs each limited to 200 iterations with varied random seeds to assess result stability. Across these five runs, the fine-tuned student model achieves a mean accuracy of 69.43% (std dev 0.17%) on the competition dataset, a 4.76 percentage-point improvement over the base model, and generalizes to 73.1% (std dev 0.18%) on the MATH-500 benchmark. We further study how response length affects answer quality across six reasoning levels (R1-R6): accuracy declines consistently from 69.43% at R1 (mean 220 words) to 41.9% at R6 (mean 31.2 words), with the two-person speed section most sensitive to token reduction. These results demonstrate that CoT distillation improves compact student models and that response length is a critical factor in mathematical reasoning quality.

Gaurab Baral, Aaditya Khanal, Yangyang Tao +1

Northern Kentucky University, Highland Heights, KY 41099, USA

CoT DistillationMathematical ReasoningKnowledge DistillationLarge Reasoning Models
AskPDF
Jun 29, 2026

Staged Hybridisation for Visual Quantum Reinforcement Learning via Knowledge Distillation

Visual environments are a demanding setting for quantum reinforcement learning (QRL): high-dimensional observations, unstable RL optimisation, and constrained variational quantum circuits (VQCs) are difficult to train jointly. This paper studies knowledge distillation (KD) as a staged hybridisation strategy for visual QRL. Instead of training a hybrid visual agent end-to-end from pixels, we first train a classical visual teacher, freeze its encoder as a feature interface, and distil the teacher's policy behaviour into compact downstream heads. These heads can be classical or VQC-based, enabling small quantum-compatible students to be evaluated under the same frozen representation as compact classical controls. We evaluate the pipeline on CartPole Pixels and Acrobot Pixels. The results show that staged KD enables shallow VQC heads to acquire non-trivial visual-control behaviour in settings where direct pixel-based training would be substantially more difficult. Angle-encoded VQC heads retain near-teacher performance, while amplitude-encoded heads push compactness to an extreme regime, at the cost of greater fragility, stronger budget sensitivity, and higher simulation time. Overall, staged KD reframes visual QRL as a compact-head learning problem, opening a practical route for training small quantum-compatible policies outside the standard end-to-end RL loop.

Javier Lazaro, Juan-Ignacio Vazquez, Pablo Garcia-Bringas

University of Deusto, Bilbao, Spain · FSAS International Quantum Center, Fujitsu

Hybrid Quantum-Classical MLVariational Quantum CircuitsVisual RLPolicy DistillationKnowledge Distillation
AskPDF
Jun 29, 2026

DuoMem: Towards Capable On-Device Memory Agents via Dual-Space Distillation

Large Language Model (LLM)-based agents can solve complex procedural tasks by interacting with environments over multiple turns, but this ability typically depends on large models, long contexts, and repeated inference calls. This makes advanced memory-augmented agents difficult to deploy on resource-constrained devices. We introduce DuoMem, a dual-space distillation framework that transfers procedural problem-solving ability from a large teacher model to compact student models. DuoMem distils in two complementary spaces: (1)context-space distillation, which replaces student-generated memories with higher-quality teacher-generated procedural memories prepended to the student's input, and (2)parameter-space distillation, which fine-tunes lightweight LoRA adapters on successful teacher trajectories. Evaluated on ALFWorld, a challenging embodied decision-making benchmark, DuoMem boosts a 4B-parameter model from 4.3% to 77.9% task success rate, closing most of the gap to a 72B teacher model (87.1%), while adding fewer than 10M trainable parameters and only a few megabytes of pre-computed teacher memories. Moreover, the DuoMem-enhanced 4B model completes tasks over 3x faster than the 72B teacher in wall-clock time, making it viable for real-time edge deployment, which would be challenging for the teacher.Extensive ablations across eight models spanning 2B-72B parameters reveal that both distillation axes contribute complementary

Peyman Hosseini, Ondrej Bohdal, Ahmed Alajrami +6

Samsung R&D Institute UK · Queen Mary University of London

LLM AgentsLLM Agent TrainingContext DistillationKnowledge DistillationMemory-Augmented Agents
AskPDF
Jun 29, 2026

ARKD: Adaptive Reinforcement Learning-Guided Bidirectional KL Divergence Distillation for Text Generation

Knowledge distillation (KD) is a key technique for compressing Large Language Models (LLMs), yet methods relying on a single KL objective often fail to balance primary distribution fitting with long-tail probability modeling, limiting both generation quality and generalization. To address this, we analyze the complementary roles of forward and reverse KL divergence (FKL/RKL) in distribution alignment from theoretical and empirical perspectives. We then propose a reinforcement-learning-based adaptive KL-weighted distillation framework, in which a policy network dynamically assigns weights to FKL and RKL based on teacher-student distributional characteristics, guided by immediate reward signals to achieve dual alignment on principal and long-tail modes. Extensive experiments demonstrate consistent improvements across Rouge-L and BertScore metrics, surpassing greedy heuristics by 0.4-0.6 points and outperforming other baseline methods on diverse benchmarks.

Zilong Liu, Xuewen Zhang, Jinrui Xing +3

Li Auto, Beijing, China · School of Software and Microelectronics, Peking University, China

RL for Knowledge DistillationLanguage Model DistillationKnowledge DistillationText GenerationAdaptive Knowledge Distillation
AskPDF
Jun 29, 2026

KbSD: Knowledge Boundary aware Self-Distillation for Behavioral Calibration in Agentic Search

Agentic search equips large language models with dynamic retrieval abilities, but existing reinforcement learning methods remain limited by reward sparsity in knowledge boundary calibration -- deciding when to trust parametric memory, when to rely on retrieved evidence, and when to abstain. Binary rewards can penalize undesirable outcomes, but provide little guidance on the reasoning process required to make calibrated decisions across different knowledge states. To address this, we propose KbSD (Knowledge boundary Self-Distillation), a framework that tackles this limitation through dense token-level supervision, outcome-level sparse rewards, and quadrant-adaptive optimization. KbSD constructs a hint-augmented teacher, architecturally identical to the student, that receives explicit knowledge boundary signals -- including parametric certainty, retrieval quality, and ground-truth answers -- to generate calibrated reasoning demonstrations. This information-asymmetric self-distillation enables dense supervision without requiring a larger external model. To further account for the heterogeneous reasoning distributions across knowledge states, we introduce a quadrant-adaptive distillation objective: reverse KL for concentrated integration, forward KL for diverse refusal, and Pareto-optimal bidirectional KL for asymmetric quadrants requiring both precision and coverage. Experiments on multiple benchmarks show that KbSD consistently improves both task accuracy and hallucination mitigation over strong baselines, with the largest gains appearing in the challenging quadrants where sparse rewards are least informative.

Tao Feng, Xinke Jiang, Chao Wu

Zhejiang University, Hangzhou, China · Peking University, Beijing, China

Language Model CalibrationAgentic RAGSelf-DistillationSparse-Reward RLKnowledge Distillation
AskPDF
Jun 28, 2026

PHF: Privileged Hidden Flow for On-Policy Self-Distillation

On-policy self-distillation (OPSD) trains a reasoning model on rollouts sampled from its own policy by matching a privileged teacher that also sees verified reference solutions. Existing OPSD objectives supervise only the output distribution, so privileged context affects training through a token-level divergence without directly supervising the internal computation that produced that distribution. We propose Privileged Hidden Flow (PHF), which additionally distills how a privileged teacher's hidden states move along the same rollout. Rather than forcing each student hidden vector to match the teacher vector at the same token position, PHF aligns token-to-token transition directions and trajectory geometry over selected generated positions. The all-layer recipe also includes an adjacent-layer relation computed from these same transitions, without pointwise hidden-state imitation. Under the same 100-step training schedule, PHF improves the Average@12 aggregate over our reproduced OPSD baseline on Qwen3-1.7B, 4B, and 8B, with observed gains of about +2.2, +1.5, and +1.7 points. The transport objective is exactly invariant to shared trajectory offsets; its local geometry term is also invariant to orthogonal transformations of transition directions. Ablations distinguish the fixed PHF recipe from pointwise hidden-state matching, single-channel transition losses, and layer-subset choices, supporting PHF as a compact hidden-flow extension to OPSD.

Yuhan Li, Mingxu Zhang, Dazhong Shen +1

1The Hong Kong University of Science and Technology (Guangzhou) · 2Nanjing University of Aeronautics and Astronautics · 3The 63rd Research Institute, National University of Defense Technology, Nanjing

On-Policy Self-DistillationLanguage Model DistillationRelational Knowledge DistillationKnowledge Distillation
AskPDF
Jun 28, 2026

DistilledGemma: Balanced Efficiency-Accuracy for Person-Place Relation Extraction from Multilingual Historical Articles

We present DistilledGemma, an efficient and accurate system for the HIPE-2026 shared task on person-place relation extraction from multilingual historical newspaper articles in English, German, and French. Our approach adopts a three-stage knowledge distillation pipeline designed to balance classification accuracy with computational efficiency. In the first stage, we systematically explored prompt engineering strategies across eight large language models to identify the most effective reasoning architecture for this challenging task. In the second stage, we applied supervised fine-tuning (SFT) via QLoRA to a Gemma 4 26B A4B teacher model, leveraging its strong multilingual capabilities to generate silver-standard chain-of-thought traces across the training corpus. In the final stage, we performed response-level distillation to transfer these learned reasoning patterns into a compact Gemma 4 E2B student model. In the official evaluation, our team WHEREAMI ranked 3rd on the standard test set with an accuracy profile mean score of 0.688, and 2nd on the binary test set with a mean score of 0.8156. Notably, by distilling knowledge from the 26B teacher to the 2.3B student, we preserved strong reasoning capabilities while reducing the deployed model size to approximately 2.3B effective parameters; the LoRA adapters used during training were merged into the student for inference. This configuration ranked 2nd in the balanced efficiency-accuracy profile across both the standard and binary test sets. These results demonstrate that knowledge distillation provides a practical and scalable solution for historical document processing, achieving competitive performance without excessive computational cost.

Youssef Aboelwafa, Ahmed Samir, Nagwa Elmakky +1

1Alexandria University, Egypt

Relation ExtractionHistorical Document AnalysisLLM Information ExtractionLanguage Model DistillationKnowledge DistillationHistorical Document Processing
AskPDF
Jun 27, 2026

Labeling Training Data for Entity Matching Using Large Language Models

Large language models (LLMs) achieve strong entity matching performance without task-specific training data, but applying them to large sets of candidate pairs is slow and costly. Matchers built on pretrained language models (PLMs), such as BERT, offer faster inference but require training data. We systematically study knowledge-distillation workflows in which an LLM teacher labels training pairs for a smaller student matcher. We vary pair selection, labeling budget, teacher model, correspondence post-processing, and student model across eight benchmarks, including unseen entities and non-English data. We compare students trained on machine-labeled data with matchers trained on the original benchmark training sets. In most cases, PLM-based matchers trained on LLM-labeled data perform similarly to those trained on benchmark sets. Pair selection matters most for small labeling budgets, where active learning is often most effective. An open-weight teacher trains competitive students, so distillation requires no closed-weight models. Compact PLM-based students compete with much larger LLM students on most tasks while requiring 34 to 459 times less inference time than direct LLM matching. On the two benchmarks with high shares of unseen products, PLM-based students substantially underperform their teachers, as do students trained on benchmark data. Under GPT-5.2 pricing, LLM labeling costs per training set average $5.86 to $8.11. These findings support knowledge distillation as a practical approach to reduce the effort of labeling task-specific training data while enabling efficient inference.

Aaron Steiner, Christian Bizer

Data and Web Science Group University of Mannheim Mannheim, Germany

LLM-Assisted AnnotationKnowledge Distillation
AskPDF