Image_Classification
-
Causal Concept Explanations for Deep Neural Models
2026-11-10 17:50:15Josh Rountree, Pulkit Verma, Oswin So, Chuchu Fan, Julie A. ShahAbstract
Concept-based explanations in neural models are important for the development, certification, and adoption of autonomous systems. Yet, interpretability remains correlational. Without methods to guide and quantify causal relationships between explanations, a concept explanation can be provided without being necessary or sufficient for the observed output thereby appearing correct while misleading. We introduce Causal Concept Wrapper Network (CCW-Net), the first use of interventional mediation analysis from fields such as economics, medicine, and epidemiology as a training objective to align causal effects of concept-based explanations with a model's output. We further apply mediation analysis as an evaluation tool to quantify, per concept label combination, how much of any model's output is causally attributable to each concept explanation. As a training objective, CCW-Net shapes a model's concept representations such that they are both necessary and sufficient for the outputs they are credited with and independent of other concepts' contributions. Counterfactual concept representations anchor a set of causal alignment criteria: necessity, sufficiency, and independence. Necessity drives irrelevant concept effects toward zero, sufficiency ensures relevant concept representations fully reconstruct a model's output, and independence enforces the total effect decomposes linearly across concepts. We evaluate CCW-Net across aircraft control, driving, and image classification. CCW-Net provides a step toward concept-based explanations that are not merely interpretable, but quantifiably causal, supporting safer evaluation, certification, and deployment of transparent and trustworthy neural models.
Abstract (translated)
URL
https://arxiv.org/abs/2611.14407
PDF
-
picoEmergencyNet : Attention-Enhanced Ultra-Small Deep Learning Model for Aerial Image Classification in Emergency Monitoring
2026-11-10 06:28:44Christos KyrkouAbstract
Real-time disaster recognition from aerial imagery requires deep learning models that operate under strict memory, compute, and latency constraints. While recent approaches achieve high accuracy, their computational cost limits deployment on resource-constrained UAV platforms. This work introduces picoEmergencyNet, an ultra-lightweight convolutional neural network designed for aerial image disaster classification. The proposed architecture combines depthwise dilated convolutions with efficient channel attention and channel shuffling to enable multi-scale feature extraction at minimal computational cost. picoEmergencyNet has 23k parameters and 7M FLOPs, requiring only 0.1 MB of model storage. Evaluated on the AIDER and AIDERv2 disaster datasets, it achieves up to 96.1% accuracy, matching or exceeding state-of-the-art lightweight models that are orders of magnitude larger. Deployment experiments on the NVIDIA Jetson AGX Orin show that, without hardware-specific optimizations, picoEmergencyNet achieves 258 FPS on GPU and 46 FPS on CPU with a 7x lower activation memory footprint compared to competing models. These results demonstrate that careful architectural specialization, rather than model scaling, enables highly efficient and accurate TinyML-ready disaster recognition for real-time UAV-based emergency monitoring.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13638
PDF
-
Earth Embeddings in Environmental Remote Sensing
2026-11-10 05:42:10Yuchi Ma, Jiaqi Yang, Mengyu Liang, Ivan Higuera-Mendieta, Caleb Robinson, Anthony Ortiz, Min Chen, David B. Lobell, Inbal Becker-Reshef, Juan M. Lavista FerresAbstract
Earth observations from satellite platforms have profoundly transformed research and applications across numerous domains. However, uneven observation coverage and frequent data gaps continue to limit their ability to capture Earth's surface dynamics. Moreover, processing multi-modal Earth observations requires substantial computational resources and sophisticated data harmonization frameworks, which may not be readily accessible to many remote sensing practitioners. Earth embeddings have emerged as promising solutions to these challenges. As general-purpose representations converted from multi-source Earth observations, Earth embeddings can support a wide range of downstream tasks with no requirement of task-specific feature design and preprocessing. In this study, we first introduce the theoretical foundations of Earth embeddings. Then, we review several representative Earth embedding products. We next examine the applications of Earth embeddings across six major domains of environmental remote sensing. Finally, we systematically assess the performance of Earth embeddings in four downstream tasks, including hyperspectral image classification, individual tree crown segmentation, crop yield prediction, and aboveground biomass mapping. Our review and case studies suggest that, despite several advantages over conventional Earth observation data, current Earth embeddings do not always outperform remote sensing imagery or hand-crafted features. Earth embeddings should not be viewed as a direct replacement for conventional EO data, but rather as representations that provide stable information across space and time. Remote sensing practitioners may consider evaluating Earth embeddings' suitability for specific tasks and using them strategically, rather than assuming superior performance by default.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13612
PDF
-
SpatialSR: Spatial Resolution-Centric Pipeline for Arbitrary-Scale Remote Sensing Super-Resolution
2026-11-09 12:47:34Yan Zhang, Runfeng Luo, Liangke Gao, Lin Yuan, Bin Xiao, Xinbo GaoAbstract
Although remote sensing image arbitrary-scale super-resolution (RSI-ASSR) methods support reconstruction at arbitrary or continuous scales, they predominantly condition on the relative scale factor and have not sufficiently investigated the absolute spatial resolution itself as a reconstruction prior, thereby disregarding the inherent spatial context of objects, which frequently leads to spatial-related distortions and compromised performance in downstream remote sensing applications. To address these challenges, we introduce SpatialSR, a SRe-centric arbitrary-scale RSISR method that leverages the spatial resolution (SRe) of remote sensing images (RSI) as an essential prior for super-resolution reconstruction. Distinct from existing SR methods, SpatialSR aims to restore high-resolution remote sensing images with a target SRe (typically set to 0.1m/pixel to align with the metadata of commonly used downstream task datasets). Specifically, we design a Spatial Resolution Estimation Module (SREM) to estimate the SRe of low-resolution (LR) images, thereby dynamically determining the required arbitrary scaling factor. Next, we employ a pretrained SRe-to-image diffusion model to extract SRe-specified texture-style codes, which are further incorporated into a carefully designed Resolution-aware Style Consistency Module (RSCM) to constrain the SRe-aware texture consistency of reconstructed images. Furthermore, to overcome the high computational expense inherent in diffusion-based SR methods, we wrap the iterative diffusion process as a Trajectory Consistency (TC) loss, which enables comparable reconstructed texture quality of the generative model in SpatialSR within a single step. Extensive experiments demonstrate that the proposed SpatialSR achieves state-of-the-art (SOTA) performance on commonly used RSISR and downstream (image classification and segmentation tasks) datasets.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13046
PDF
-
LEF-JEPA: Denoising-Trained Latent Energy Fields for Spatial Representation Learning
2026-11-06 05:00:20Zhiyuan Liu, Zhihao Xu, Sicheng Liu, Huanhuan HuangAbstract
A joint-embedding predictive architecture (JEPA) that applies prediction supervision only at masked positions struggles to clearly capture object shapes and boundaries in its latent representations. We further observe that some existing methods improve spatial structure while also retaining rich variations in texture and color. However, preserving more low-level visual information does not necessarily make representations better suited to prediction and action. This raises a key question: how can we strengthen the spatial structure in latent representations that supports prediction and action without over-encoding these details? We introduce LEF-JEPA, which learns spatial structure through target-guided latent recovery and transfers this supervision to the visual encoder through a relative-energy constraint; the energy module is used only during pretraining and adds no inference overhead. To reduce the cost of energy optimization during pretraining, we further construct a conditional Ising Hamiltonian that allows low-energy-state search to be performed by specialized Ising solvers. On physical hardware, we validate the feasibility of achieving lower forward latency with fewer local updates. Under a unified CIFAR-10 pretraining and frozen-encoder evaluation protocol, both realizations outperform V-JEPA2.1 in image classification, foreground segmentation, and visual planning.
Abstract (translated)
URL
https://arxiv.org/abs/2611.08989
PDF
-
Improving Context-Shift Robustness of Convolutional Networks via Context-Regularized Cross-Entropy
2026-11-05 15:16:49Jinen Setpal, Chaoyue LiuAbstract
Image classification models often encounter objects in diverse and unpredictable contexts at test time, which may differ substantially from those seen during training. This phenomenon, termed context shift, exposes a key fragility in neural networks: when models rely on non-causal contextual cues for prediction, shifts in context can lead to significant performance degradation. In this paper, we propose a novel training framework to improve the robustness of convolutional neural networks under context shift. Our approach leverages spatial feature attribution to guide models toward making predictions that rely less on contextual regions and more on object-relevant features. As a first step, recognizing a theoretical limitation of existing feature attribution methods we introduce a new variant, ContrastiveCAMs, which produces more faithful attribution maps of model predictions. Building on ContrastiveCAMs, we further propose Context-Regularized Cross-Entropy (CR-CE), a modification of the standard cross-entropy loss that regularizes the model's attention by suppressing the influence of contextual regions, thereby improving robustness to context shift. We evaluate the effectiveness of our approach on several medium-to-large scale datasets (Waterbirds, Spawrious, ImageNet/ImageNet-BG, Hard-ImageNet), and report consistent improvements in context-shift robustness.
Abstract (translated)
URL
https://arxiv.org/abs/2611.08157
PDF
-
Toward an Optimal Configuration for Tree Species Classification from Bark Images: A Leakage-Safe Ablation Study
2026-11-03 22:05:57Maeve C\^ot\'eAbstract
Background: The automatic identification of tree species from bark photographs is of practical interest for forestry inventory and the wood-processing industry, yet the design choices that govern the accuracy of deep-learning approaches have not been studied systematically, and some prior results may be inflated by leakage of images of the same trunk across training and test partitions. Methods: A systematic ablation study is conducted on the public BarkNet 1.0 dataset (20 species) under the established tree-level, leakage-safe partitioning scheme. Using ConvNeXt-V2 backbones, the following factors are varied in turn and evaluated by five-fold cross-validation: patch size, the strategy by which per-patch predictions are aggregated into an image-level decision, the training regime (single- versus two-stage), backbone capacity, the whole-image baseline, and the number of patches available at inference. Results: Per-patch accuracy rises monotonically with patch size (from 75.6% at 160 px to 89.6% at 512 px), and image-level accuracy rises in step, plateauing near 94.9% around 512 px. Learned aggregation exceeds majority voting by at most 0.7 pp, an advantage that shrinks as patches grow and attention becomes near-uniform. Backbone capacity (pico to tiny) alters accuracy by less than a point, and two-stage training offers no advantage over one-stage. Whole-image classification (91.9%) trails patch-based aggregation, since it admits no bag to average over; accuracy increases with the number of patches at inference, saturating around 64. Conclusions: The study yields a simple recommended configuration: patches as large as possible while leaving enough to average over (around 512 px here), with a nano backbone. It also quantifies the second-order role of learned aggregation and model capacity, providing a reproducible reference point for practitioners and future research
Abstract (translated)
URL
https://arxiv.org/abs/2611.05680
PDF
-
IGDA-ICMH: an Importance Guided Feature and Network Dual Adaptation Framework towards Image Compression for Machine and Human Vision
2026-10-30 15:54:03Yanbo Gao, Xinyu Zhao, Shuai Li, Xun Cai, Mao Ye, Li LiAbstract
With the surge of machine vision applications, image compression for both machine and human vision (ICMH) has become an important research area. Existing studies mainly focus on task-specific adaptation or multi-task joint optimization, yet they often overlook the unequal contributions of different features and network layers to downstream machine vision tasks. To address this issue, we first quantitatively investigate the impact of different feature and network layers on machine vision performance, providing a principled understanding of their importance. Based on this analysis, an Importance Guided feature and network Dual Adaptation (IGDA) framework is developed for ICMH. It adopts implicit neural representation as the basic architecture and designs an efficient adaptation structure with both feature and network weight adaptation. An Importance-Guided Network Adaptation (IGNA) module is developed with a low-rank adaptation strategy to selectively adjust key network layers according to their importance in different tasks. An Importance-Guided Feature Adaptation (IGFA) module is developed with a low-rank and Kronecker product-based transform to explore the non-local property of the features in order to reduce the number of adaptation parameters. Extensive experiments on representative machine vision tasks, including image classification and object detection, demonstrate that our IGDA-ICMH framework achieves significantly better performance than the existing implicit neural representation-based methods for ICMH.
Abstract (translated)
URL
https://arxiv.org/abs/2610.41924
PDF
-
Breaking Information Islands in Sparse Tuning via Small-World Connectivity
2026-10-30 12:35:08Xianchao Guan, Zijun Xiong, Yifeng Wang, Yubo Cui, Yaowei Wang, Guo-Sen Xie, Xin Li, Zheng ZhangAbstract
Sparse tuning is widely used to adapt large language models due to its parameter efficiency. However, its effectiveness depends not only on the number of trainable parameters, but also on the feature-interaction topology of the sparse update's structural support. Limited or uneven connectivity in this support may isolate subsets of feature dimensions, forming information islands that restrict direct task-specific interactions through the trainable update. To address this issue, we propose Small-World Induced Fine-Tuning (SWIFT), a sparse tuning framework inspired by small-world connectivity. SWIFT combines block-diagonal sparse updates for local feature mixing with a fixed deterministic scattering permutation that routes the mixed outputs across feature groups without additional trainable parameters. We further show theoretically that certain random sparse-update models and block-diagonal updates can exhibit information islands, whereas a sufficiently scattering permutation guarantees positive block-level expansion when the trainable block-diagonal factor consists of equal-sized, structurally dense square blocks. Extensive experiments on commonsense reasoning, natural language generation, and image classification demonstrate that SWIFT outperforms the evaluated parameter-efficient fine-tuning (PEFT) baselines on average and can match or surpass full fine-tuning. Implementation is available at this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2610.41665
PDF
-
With foundation-model features, aggregator complexity does not improve cross-hospital discrimination in gastric whole-slide image classification
2026-10-30 06:43:47Nheng Vanchhay, Ali Hussain, Khan Ziaullah, Moon-Il Joo, Md Ariful Islam Mozumder, Rashedul Islam Sumon, Won-Ae Lee, Min-Sun Cho, Hyeong-Chan Shin, Yujun Park, Baek Hwan Cho, Nam-Hoon Cho, Heejung An, Heung-Kook Choi, Haeyoun Kang, Hyunki Kim, Hee Cheol KimAbstract
Whole-slide image classifiers are usually developed within one hospital and often go untested at unseen hospitals. We assembled 5,036 haematoxylin-and-eosin-stained gastric slides from five Korean hospitals, scanned at 20x and 40x, and compared multiple-instance learning aggregators under leave-one-hospital-out validation on frozen pathology foundation-model features. A hierarchical patch-region-slide graph aggregator (CHIME-MIL) with site alignment and two-scale fusion reached a mean area under the receiver operating characteristic curve (AUC) of 0.837, above six conventional attention and transformer aggregators trained without these steps. Under a matched recipe it did not exceed attention-based pooling, and global max pooling given the same alignment and fusion was better (AUC 0.855, P = 0.0002). The alignment, which subtracts each hospital's mean feature vector without labels, onboarded a new hospital from ten unlabelled slides and cost 0.002 AUC without held-out-hospital data, but its accuracy gains were small and inconsistent. The model had low top-label calibration error (0.053) but under-confident malignancy probabilities, and removing its most-attended patches lowered the malignancy probability far more than removing random patches. With these features, more complex aggregators did not improve discrimination; what a deployable system adds is label-free onboarding, calibration checks and auditable evidence.
Abstract (translated)
URL
https://arxiv.org/abs/2610.41328
PDF
-
Beyond CLIP: Calibrated Defect Detection With System One Decision Models
2026-10-30 03:21:02Fadel M. Megahed, Ying-Ju ChenAbstract
Cameras and vision sensors are widely used for product and process inspection in manufacturing environments. Traditionally, this process required thousands of images of both nominal and defective products to train such models. Recent research shows that vision transformer (ViT) models, such as CLIP, can reduce this requirement to 50-100 images per nominal and defective class. However, collecting even 50 defective images can be difficult in high-yield or low-volume manufacturing processes. In this paper, we examine the utility of two recently proposed structured decision models (Jev and Clef-Flash) in applications using as few as k=2 reference images per class at inference. In the process, we provide three main contributions: (a) a framework for k-shot learning of these models for image classification/inspection applications; (b) an agentic approach for automatically refining the text descriptions that guide image classification in these decision models; and (c) an approach to calibrate the resulting defect probabilities to improve their reliability for inspection decisions. We test these contributions in five manufacturing case studies. Using two reference images per class, both Jev and Clef-Flash outperform CLIP's test accuracies with the same references (0.75-1.00 [Jev] and 0.71-0.99 [Clef-Flash] vs 0.67-0.84 [CLIP]). Furthermore, we show that our agentic refinement of text descriptions significantly improves accuracy across 6 of 8 model-application combinations, and our calibration approach reduces the Brier score in 13 of the 15 model-application configurations. To support the reproducibility of our work, we provide our replication code and have created the R "decisions" package for evaluating additional models. We also provide a publicly available web app that allows practitioners to inspect their own images with no code.
Abstract (translated)
URL
https://arxiv.org/abs/2610.41199
PDF
-
Episodic looking - using reinforcement learning to bias perception enhances few-shot and continual learning with protection from forgetting
2026-10-28 10:09:11David Rawlinson, Gideon KowadloAbstract
There is a fundamental contradiction between the desire for models that can generalize, and models that can continuously adapt to recent input with changing statistics. These conditions, captured in few-shot and continual learning settings, can bias or disrupt learned representations. This harms generalization to a broader context, leaving the model unable to overcome input variance. This paper proposes a solution based on the Complementary Learning System architecture (CLS), in which a Short-Term Memory (STM) delegates generalization to a slow, statistical Long-Term Memory (LTM). Delegation is achieved by working in a space of robust, generalized perceptions produced by the LTM. With this separation, the STM can focus on memorizing recent input and avoiding interference between memories. We show that on an image classification task, the resulting CLS model is more resistant to forgetting in a domain-incremental continual learning setting than a baseline ResNet LTM, and generalizes more effectively in a few-shot setting than other continual learners. In the continual setting, overall accuracy is slightly worse than existing continual learning models, but with a different mix of performance characteristics. Results analysis suggests that the model is limited by the representations and representational plasticity of the LTM. In future work, we will seek to address this by replaying STM memories to the LTM.
Abstract (translated)
URL
https://arxiv.org/abs/2610.38577
PDF
-
Measuring Deep Learning Performance - An Empirical Study of Performance Distributions Across Architectures and Tasks
2026-10-27 22:47:10Kevin L Coakley, Odd Erik GundersenAbstract
Context: Non-determinism in deep learning algorithm design and implementation leads to performance variation, meaning model performance is not a single value, but rather a distribution. These model performance distributions are underexplored despite their impact on robustness. Objective: We investigate the robustness of deep learning performance to sources of non-determinism, specifically focusing on how performance distributions differ across various architectures and tasks. Method: We conducted 186 experiments on image classification and time series forecasting architectures. Each experiment was run 100 times with different random seeds to generate performance distributions, resulting in 18,600 runs. Robustness was quantified using metrics for spread, symmetry, and tail risk. Results: Performance distributions are frequently non-Gaussian, particularly in time series forecasting. Model size does not systematically affect robustness -- larger image classification models show fewer outliers but not lower spread, while smaller time series models show lower spread but more extreme underperformers. Training duration does not scale linearly; early stopping effectively balances performance and robustness. Mean performance does not predict robustness - time series forecasting shows moderate correlation while image classification shows none. Time series models produce nearly three times more underperforming outliers than image classification models, indicating substantially higher tail risk. Conclusions: Tail risk poses serious concerns for Trustworthy AI in high-stakes applications. Models performing well on average may exhibit long tails and extreme outliers revealed only through distributional analysis. Mean performance alone should not guide model selection; assessment of spread, symmetry, and tail risk is essential for reliable model assessment where consistent performance is critical.
Abstract (translated)
URL
https://arxiv.org/abs/2610.38037
PDF
-
RePAIR: Inverting Backdoor Attacks in Semantic Segmentation
2026-10-27 10:53:47Arnold Brosch, Kira MaagAbstract
Despite significant advances in computer vision, semantic segmentation models remain vulnerable to backdoor attacks, posing a security risk. These attacks redirect pixels of a chosen victim class to an attacker-specified target class when a trigger is present, while preserving clean predictions. Existing defenses are largely derived from image classification, while dedicated segmentation defenses remain scarce and typically require retraining or model modification. We introduce RePAIR, a post-hoc, label-free, attack-unaware defense based on the observation that backdoor-induced shifts can be stronger in the classifier head than in internal representations, leaving recoverable victim-class semantics. RePAIR infers the victim-target relation from an unlabeled, potentially contaminated image pool, constructs a safety-approved semantic reference, and localizes suspicious pixels by comparing head predictions with internal semantic evidence. It then applies feature-space and output corrections toward the inferred victim semantics, while conservative localization and output constraints limit changes to benign predictions. We model the attack as adversarial probability transport and repair as its approximate reversal, deriving bounds and sufficient conditions for localization, residual attack success, clean-side changes, defense-induced target predictions, and attack-unaware pair identification. Experiments across segmentation architectures and attack settings show that RePAIR substantially reduces attack success while largely preserving clean performance.
Abstract (translated)
URL
https://arxiv.org/abs/2610.37015
PDF
-
SAP-ViT: A Self-Attributing Patch Vision Transformer for Medical Image Classification
2026-10-26 15:19:58Wei Zhang, Wolfgang Nejdl, Ming TangAbstract
Vision Transformer (ViT) classifiers commonly rely on a global classification (CLS) token for image-level prediction, which is poorly aligned with local interpretability requirements in medical imaging. We propose SAP-ViT, a Self-Attributing Patch Vision Transformer that replaces CLS-based classification with direct aggregation of last-layer patch embeddings. This design makes class predictions decomposable into patch-level contributions as part of the forward pass. To encourage concise and readable evidence, SAP-ViT applies sparsity regularization to the patch-weight distribution. Since self-attention mixes information across patch tokens, we further introduce an input-space back-mapping mechanism to estimate which input patches contribute to influential output patch embeddings. We evaluate SAP-ViT qualitatively and quantitatively on multiple medical image classification datasets spanning different imaging modalities. The results show that SAP-ViT maintains strong classification performance while providing consistent and interpretable patch-level explanations across datasets. Ablation results further show that both sparsity regularization and the alignment component of input-space back-mapping contribute to the final explanation quality. Overall, SAP-ViT provides a favorable trade-off between predictive accuracy and interpretability for ViT-based medical image classification.
Abstract (translated)
URL
https://arxiv.org/abs/2610.35593
PDF
-
S$^2$-RL: Sample-Set Dual Reinforcement Learning for Generative Semantic Segmentation Dataset Distillation
2026-10-26 14:21:31Haoyu Wang, Fei Zhou, Qingqing Qiu, Lei Zhang, Wei Wei, Chen DingAbstract
While dataset distillation has witnessed significant advances in image classification, its extension to dense prediction tasks such as semantic segmentation is plagued by two core bottlenecks: bi-level optimization-based approaches suffer from prohibitive computational costs stemming from pixel-wise gradient unrolling, whereas proxy-based generative methods merely infuse visual textures into fixed semantic masks derived from real data, thus failing to condense rich semantic knowledge into a compact set of informative, novel scene generations. To address both bottlenecks simultaneously, we propose a novel Sample-Set Reinforcement Learning (S$^2$-RL) framework for generative semantic segmentation dataset distillation (SSDD). Specifically, S$^2$-RL formulates SSDD as a diffusion model based text-to-image generation paradigm, enabling flexible generation of images with informative, novel semantic distributions via a concise text prompt. Furthermore, we fine-tune a diffusion policy via Group Relative Policy Optimization (GRPO), leveraging a sample-set dual reward paradigm: for individual samples, it enforces semantic alignment with input text prompts and intra-class feature diversity; for sample sets, it maximizes inter-sample semantic distribution diversity by solving a contextual multi-armed bandit problem. These design choices enable S$^2$-RL to distill large-scale semantic segmentation datasets into a compact set of generated samples that encapsulate diverse, informative semantic knowledge without the need of pixel-wise gradient unrolling. Extensive experiments on the ADE20K and COCO datasets demonstrate that S$^2$-RL achieves substantial and consistent improvements over state-of-the-art baselines, establishing a new benchmark for SSDD.
Abstract (translated)
URL
https://arxiv.org/abs/2610.35491
PDF
-
ACSAN: All-Convolutional Self-Attention Network Integrating Convolutional Operations with Vision Transformer
2026-10-26 12:58:25Zimian Wei, Tianqi Zhang, Hengyue PanAbstract
In the past few years, Vision Transformer (ViT) has achieved excellent performance in the field of computer vision. However, there still some problems remain. For instance, ViTs heavily rely on large-scale training data since they simply contain a large number of trainable parameters, which slows down their learning efficiency and limits their applicability in resource-constrained scenarios. In this paper, we propose a novel deep learning model, namely All-Convolutional Self-Attention Network (ACSAN), which applies convolution operations to replace fully-connected matrices multiplications in the ViT framework. To implement ACSAN, we firstly design Dynamic Convolutional Self-Attention (DCSA) as the basic module, which employs multi-scale dilated convolutions to generate Query, Key, and Value. We defined spatial self-attention and channel self-attention to generate attention features respectively. The two attention features are finally merged as the output of the DCSA module. Based on DCSA, we then construct the basic ACSAN block. Comparing with ViT, we use DCSA to replace the regular self-attention operations, and use convolution layers to replace MLPs. To evaluate the performance of the proposed method, we train ACSAN on three widely-used image classification databases, i.e. CIFAR-10, CIFAR-100 and ImageNet-1k, from scratch. Experimental results show that ACSAN achieves excellent classification accuracy comparing with standard ViT with comparable or fewer parameters, as well as other state-of-the-art methods. Ablation studies verify that each component contributes positively to the final performance.
Abstract (translated)
URL
https://arxiv.org/abs/2610.35375
PDF
-
SVD-Attention: Efficiently Analyzing Attention Blocks in Transformers
2026-10-26 07:40:43Debopam Sanyal, Hongjie Chen, Riya Elizabeth Kuriyan, Arth Tak, Divya Kothandaraman, Akshay Mehra, Joshua Kimball, Alexey TumanovAbstract
Pretrained transformers spend a large share of their inference compute inside self-attention, yet the per-head query-key interaction is structurally low-rank by construction. We ask a diagnostic question: across layers and architectures, which heads actually use the full rank budget on real data, and which do not? We propose SVD-Attention, a post-hoc method that decomposes the per-head interaction matrix $M_h$ and scores every singular mode by both its spectral magnitude and its activation variance across tokens on a small calibration set. The score keeps the modes the head can amplify and that real data actually activates, and discards the rest; the resulting factorization is exact for the retained modes and needs no finetuning. Across seven pretrained models and five task types, the diagnostic finds a task-dependent pattern: redundancy sits in the early to middle layers for sentiment and image classification and for generation, and in the late layers for dense prediction and extractive question answering. It is also predictive: truncating the layers it flags costs at most $1.75$ F1, while truncating the same number of layers elsewhere in the same model costs up to $50$ F1. SVD-Attention is an analysis tool rather than a compression method: acting on it reduces inference FLOPs by $0.8$-$1.5\%$ on its own and by $8$-$11\%$ when composed with weight pruning, with $0$-$4\%$ change in task quality.
Abstract (translated)
URL
https://arxiv.org/abs/2610.34971
PDF
-
Greedy Local Learning for Language Model Pretraining: Gaps and Objective Design
2026-10-26 05:00:35Jihwan Moon, Sheir A. Zaheer, Jinmyoung Lee, Gunhee Kim, Chan Y. ParkAbstract
Greedy block-wise local learning splits a network into gradient-isolated blocks trained by local auxiliary losses, deleting the backward pass between blocks: inter-stage communication becomes forward-only and every block can step its optimizer independently, properties directly relevant to decentralized model-parallel training. Local learning is competitive with end-to-end backpropagation on image classification, and on small Transformers it is known to trade a worse best loss for parallel speedup. How this loss gap behaves in autoregressive language model (LM) pretraining at larger scale, and which auxiliary designs reduce it, has not been measured. We present a token-budget-matched empirical study at 125M and 400M parameters with $K \in \{1,2,4\}$ blocks at Chinchilla-optimal budgets, factorizing the auxiliary design into network architecture and training objective. We observe: (i) the gap to end-to-end training more than doubles from $K=2$ to $K=4$, but at $K=4$ shrinks from 125M to 400M; (ii) replacing an MLP auxiliary with a Transformer-based one is a strong network-side intervention, recovering 22-41% of the gap; (iii) a multi-token-prediction (MTP) auxiliary objective helps at the first block boundary, whereas adding it at deeper boundaries hurts, and restricting it to the first block yields the best $K=4$ configuration ($+0.062$ vs. $+0.075$ nats at 400M); and (iv) deployment-style per-block execution reduces activation memory by up to $2.2\times$. We frame these results as an empirically grounded method direction rather than a finalized method: local objectives should apply future-predictive pressure selectively across boundaries while resisting shortcuts that bypass predictive content.
Abstract (translated)
URL
https://arxiv.org/abs/2610.04867
PDF
-
AutoMIL: Automated Multiple Instance Learning for Whole-Slide Image Classification
2026-10-23 10:17:42Jonas Waibel, Frank Kramer, Dominik M\"ullerAbstract
Developing multiple instance learning pipelines for the classification of whole-slide images involves many interdependent design decisions and expertise in both pathology and machine learning. We present AutoMIL, an open-source framework that applies principles from the field of automated machine learning to tackle this task by automating integral pipeline steps, such as preprocessing, training and evaluation. In a benchmark evaluation with default settings, AutoMIL reached ROC-AUC scores of 0.94 on UBC-OCEAN, 0.99 on PCam and 0.93-0.96 on a TCGA-NSCLC subset, comparable to specialised pipelines in related work. Ensembling mainly improved performance on imbalanced data, showing that an automated pipeline must adapt to dataset characteristics. AutoMIL thus provides an accessible baseline for whole-slide image classification without manual pipeline configuration.
Abstract (translated)
URL
https://arxiv.org/abs/2610.32452
PDF
