Current artificial neural networks are trained with parameters encoded as floating point numbers that occupy lots of memory space at inference time. Due to the increase in the size of deep learning models, it is becoming very difficult to consider training and using artificial neural networks on edge devices. Binary neural networks promise to reduce the size of deep neural network models, as well as to increase inference speed while decreasing energy consumption. Thus, they may allow the deployment of more powerful models on edge devices. However, binary neural networks are still proven to be difficult to train using the backpropagation-based gradient descent scheme. This paper extends the work of \cite{crulis2023alternatives}, which proposed adapting to binary neural networks two promising alternatives to backpropagation originally designed for continuous neural networks, and experimented with them on simple image classification datasets. This paper proposes new experiments on the ImageNette dataset, compares three different model architectures for image classification, and adds two additional alternatives to backpropagation.
While the traditional formulation of machine learning tasks is in terms of performance on average, in practice we are often interested in how well a trained model performs on rare or difficult data points at test time. To achieve more robust and balanced generalization, methods applying sharpness-aware minimization to a subset of worst-case examples have proven successful for image classification tasks, but only using overparameterized neural networks under which the relative difference between "easy" and "hard" data points becomes negligible. In this work, we show how such a strategy can dramatically break down under simpler models where the difficulty gap becomes more extreme. As a more flexible alternative, instead of typical sharpness, we propose and evaluate a training criterion which penalizes poor loss concentration, which can be easily combined with loss transformations such exponential tilting, conditional value-at-risk (CVaR), or distributionally robust optimization (DRO) that control tail emphasis.
Vision Transformers (ViTs) mark a revolutionary advance in neural networks with their token mixer's powerful global context capability. However, the pairwise token affinity and complex matrix operations limit its deployment on resource-constrained scenarios and real-time applications, such as mobile devices, although considerable efforts have been made in previous works. In this paper, we introduce CAS-ViT: Convolutional Additive Self-attention Vision Transformers, to achieve a balance between efficiency and performance in mobile applications. Firstly, we argue that the capability of token mixers to obtain global contextual information hinges on multiple information interactions, such as spatial and channel domains. Subsequently, we propose Convolutional Additive Token Mixer (CATM) employing underlying spatial and channel attention as novel interaction forms. This module eliminates troublesome complex operations such as matrix multiplication and Softmax. We introduce Convolutional Additive Self-attention(CAS) block hybrid architecture and utilize CATM for each block. And further, we build a family of lightweight networks, which can be easily extended to various downstream tasks. Finally, we evaluate CAS-ViT across a variety of vision tasks, including image classification, object detection, instance segmentation, and semantic segmentation. Our M and T model achieves 83.0\%/84.1\% top-1 with only 12M/21M parameters on ImageNet-1K. Meanwhile, throughput evaluations on GPUs, ONNX, and iPhones also demonstrate superior results compared to other state-of-the-art backbones. Extensive experiments demonstrate that our approach achieves a better balance of performance, efficient inference and easy-to-deploy. Our code and model are available at: \url{https://github.com/Tianfang-Zhang/CAS-ViT}
Lightweight and effective models are essential for devices with limited resources, such as intelligent vehicles. Structured pruning offers a promising approach to model compression and efficiency enhancement. However, existing methods often tie pruning techniques to specific model architectures or vision tasks. To address this limitation, we propose a novel unified pruning framework Comb, Prune, Distill (CPD), which addresses both model-agnostic and task-agnostic concerns simultaneously. Our framework employs a combing step to resolve hierarchical layer-wise dependency issues, enabling architecture independence. Additionally, the pruning pipeline adaptively remove parameters based on the importance scoring metrics regardless of vision tasks. To support the model in retaining its learned information, we introduce knowledge distillation during the pruning step. Extensive experiments demonstrate the generalizability of our framework, encompassing both convolutional neural network (CNN) and transformer models, as well as image classification and segmentation tasks. In image classification we achieve a speedup of up to x4.3 with a accuracy loss of 1.8% and in semantic segmentation up to x1.89 with a 5.1% loss in mIoU.
Pre-trained encoders available online have been widely adopted to build downstream machine learning (ML) services, but various attacks against these encoders also post security and privacy threats toward such a downstream ML service paradigm. We unveil a new vulnerability: the Pre-trained Encoder Inference (PEI) attack, which can extract sensitive encoder information from a targeted downstream ML service that can then be used to promote other ML attacks against the targeted service. By only providing API accesses to a targeted downstream service and a set of candidate encoders, the PEI attack can successfully infer which encoder is secretly used by the targeted service based on candidate ones. Compared with existing encoder attacks, which mainly target encoders on the upstream side, the PEI attack can compromise encoders even after they have been deployed and hidden in downstream ML services, which makes it a more realistic threat. We empirically verify the effectiveness of the PEI attack on vision encoders. we first conduct PEI attacks against two downstream services (i.e., image classification and multimodal generation), and then show how PEI attacks can facilitate other ML attacks (i.e., model stealing attacks vs. image classification models and adversarial attacks vs. multimodal generative models). Our results call for new security and privacy considerations when deploying encoders in downstream services. The code is available at https://github.com/fshp971/encoder-inference.
Recent work has uncovered alarming disparities in the performance of machine learning models in healthcare. In this study, we explore whether such disparities are present in the UK Biobank fundus retinal images by training and evaluating a disease classification model on these images. We assess possible disparities across various population groups and find substantial differences despite strong overall performance of the model. In particular, we discover unfair performance for certain assessment centres, which is surprising given the rigorous data standardisation protocol. We compare how these differences emerge and apply a range of existing bias mitigation methods to each one. A key insight is that each disparity has unique properties and responds differently to the mitigation methods. We also find that these methods are largely unable to enhance fairness, highlighting the need for better bias mitigation methods tailored to the specific type of bias.
The success of large-scale pre-trained models has established fine-tuning as a standard method for achieving significant improvements in downstream tasks. However, fine-tuning the entire parameter set of a pre-trained model is costly. Parameter-efficient transfer learning (PETL) has recently emerged as a cost-effective alternative for adapting pre-trained models to downstream tasks. Despite its advantages, the increasing model size and input resolution present challenges for PETL, as the training memory consumption is not reduced as effectively as the parameter usage. In this paper, we introduce Fine-grained Prompt Tuning plus (FPT+), a PETL method designed for high-resolution medical image classification, which significantly reduces the training memory consumption compared to other PETL methods. FPT+ performs transfer learning by training a lightweight side network and accessing pre-trained knowledge from a large pre-trained model (LPM) through fine-grained prompts and fusion modules. Specifically, we freeze the LPM of interest and construct a learnable lightweight side network. The frozen LPM processes high-resolution images to extract fine-grained features, while the side network employs corresponding down-sampled low-resolution images to minimize the memory usage. To enable the side network to leverage pre-trained knowledge, we propose fine-grained prompts and fusion modules, which collaborate to summarize information through the LPM's intermediate activations. We evaluate FPT+ on eight medical image datasets of varying sizes, modalities, and complexities. Experimental results demonstrate that FPT+ outperforms other PETL methods, using only 1.03% of the learnable parameters and 3.18% of the memory required for fine-tuning an entire ViT-B model. Our code is available https://github.com/YijinHuang/FPT.
Recent advancements in low-cost ensemble learning have demonstrated improved efficiency for image classification. However, the existing low-cost ensemble methods show relatively lower accuracy compared to conventional ensemble learning. In this paper, we propose a new low-cost ensemble learning, which can simultaneously achieve high efficiency and classification performance. A CNN is transformed into a multi-branch structure without introduction of additional components, which maintains the computational complexity as that of the original single model and also enhances diversity among the branches' outputs via sufficient separation between different pathways of the branches. In addition, we propose a new strategy that applies grouped convolution in the branches with different numbers of groups in different branches, which boosts the diversity of the branches' outputs. For training, we employ knowledge distillation using the ensemble of the outputs as the teacher signal. The high diversity among the outputs enables to form a powerful teacher, enhancing the individual branch's classification performance and consequently the overall ensemble performance. Experimental results show that our method achieves state-of-the-art classification accuracy and higher uncertainty estimation performance compared to previous low-cost ensemble methods. The code is available at https://github.com/hjdw2/SEMBG.
Recent ensemble learning methods for image classification have been shown to improve classification accuracy with low extra cost. However, they still require multiple trained models for ensemble inference, which eventually becomes a significant burden when the model size increases. In this paper, we propose a low-cost ensemble learning and inference, called Network Fission Ensembles (NFE), by converting a conventional network itself into a multi-exit structure. Starting from a given initial network, we first prune some of the weights to reduce the training burden. We then group the remaining weights into several sets and create multiple auxiliary paths using each set to construct multi-exits. We call this process Network Fission. Through this, multiple outputs can be obtained from a single network, which enables ensemble learning. Since this process simply changes the existing network structure to multi-exits without using additional networks, there is no extra computational burden for ensemble learning and inference. Moreover, by learning from multiple losses of all exits, the multi-exits improve performance via regularization, and high performance can be achieved even with increased network sparsity. With our simple yet effective method, we achieve significant improvement compared to existing ensemble methods. The code is available at https://github.com/hjdw2/NFE.
In the domain of black-box model extraction, conventional methods reliant on soft labels or surrogate datasets struggle with scaling to high-dimensional input spaces and managing the complexity of an extensive array of interrelated classes. In this work, we present a novel approach that utilizes SHAP (SHapley Additive exPlanations) to enhance synthetic data generation. SHAP quantifies the individual contributions of each input feature towards the victim model's output, facilitating the optimization of an energy-based GAN towards a desirable output. This method significantly boosts performance, achieving a 16.45% increase in the accuracy of image classification models and extending to video classification models with an average improvement of 26.11% and a maximum of 33.36% on challenging datasets such as UCF11, UCF101, Kinetics 400, Kinetics 600, and Something-Something V2. We further demonstrate the effectiveness and practical utility of our method under various scenarios, including the availability of top-k prediction probabilities, top-k prediction labels, and top-1 labels.
This paper examines the mathematical foundations of transformer architectures, highlighting their limitations particularly in handling long sequences. We explore prerequisite models such as Mamba, Vision Mamba (ViM), and LV-ViT that pave the way for our proposed architecture, DeMansia. DeMansia integrates state space models with token labeling techniques to enhance performance in image classification tasks, efficiently addressing the computational challenges posed by traditional transformers. The architecture, benchmark, and comparisons with contemporary models demonstrate DeMansia's effectiveness. The implementation of this paper is available on GitHub at https://github.com/catalpaaa/DeMansia
Counterfactual explanations (CEs) aim to enhance the interpretability of machine learning models by illustrating how alterations in input features would affect the resulting predictions. Common CE approaches require an additional model and are typically constrained to binary counterfactuals. In contrast, we propose a novel method that operates directly on the latent space of a generative model, specifically a Diffusion Autoencoder (DAE). This approach offers inherent interpretability by enabling the generation of CEs and the continuous visualization of the model's internal representation across decision boundaries. Our method leverages the DAE's ability to encode images into a semantically rich latent space in an unsupervised manner, eliminating the need for labeled data or separate feature extraction models. We show that these latent representations are helpful for medical condition classification and the ordinal regression of severity pathologies, such as vertebral compression fractures (VCF) and diabetic retinopathy (DR). Beyond binary CEs, our method supports the visualization of ordinal CEs using a linear model, providing deeper insights into the model's decision-making process and enhancing interpretability. Experiments across various medical imaging datasets demonstrate the method's advantages in interpretability and versatility. The linear manifold of the DAE's latent space allows for meaningful interpolation and manipulation, making it a powerful tool for exploring medical image properties. Our code is available at https://doi.org/10.5281/zenodo.13859266.
Recent advancements in transformers, specifically self-attention mechanisms, have significantly improved hyperspectral image (HSI) classification. However, these models often suffer from inefficiencies, as their computational complexity scales quadratically with sequence length. To address these challenges, we propose the morphological spatial mamba (SMM) and morphological spatial-spectral Mamba (SSMM) model (MorpMamba), which combines the strengths of morphological operations and the state space model framework, offering a more computationally efficient alternative to transformers. In MorpMamba, a novel token generation module first converts HSI patches into spatial-spectral tokens. These tokens are then processed through morphological operations such as erosion and dilation, utilizing depthwise separable convolutions to capture structural and shape information. A token enhancement module refines these features by dynamically adjusting the spatial and spectral tokens based on central HSI regions, ensuring effective feature fusion within each block. Subsequently, multi-head self-attention is applied to further enrich the feature representations, allowing the model to capture complex relationships and dependencies within the data. Finally, the enhanced tokens are fed into a state space module, which efficiently models the temporal evolution of the features for classification. Experimental results on widely used HSI datasets demonstrate that MorpMamba achieves superior parametric efficiency compared to traditional CNN and transformer models while maintaining high accuracy. The code will be made publicly available at \url{https://github.com/mahmad000/MorpMamba}.
Spatial-Spectral Mamba (SSM) improves computational efficiency and captures long-range dependencies, addressing Transformer limitations. However, traditional Mamba models overlook rich spectral information in HSIs and struggle with high dimensionality and sequential data. To address these issues, we propose the SSM with multi-head self-attention and token enhancement (MHSSMamba). This model integrates spectral and spatial information by enhancing spectral tokens and using multi-head attention to capture complex relationships between spectral bands and spatial locations. It also manages long-range dependencies and the sequential nature of HSI data, preserving contextual information across spectral bands. MHSSMamba achieved remarkable classification accuracies of 97.62\% on Pavia University, 96.92\% on the University of Houston, 96.85\% on Salinas, and 99.49\% on Wuhan-longKou datasets. The source code is available at \href{https://github.com/MHassaanButt/MHA\_SS\_Mamba}{GitHub}.
Multiple instance learning (MIL) has become a preferred method for gigapixel whole slide image (WSI) classification without requiring patch-level annotations. Current MIL research primarily relies on embedding-based approaches, which extract patch features using a pre-trained feature extractor and aggregate them for slide-level prediction. Despite the critical role of feature extraction, there is limited guidance on selecting optimal feature extractors to maximize WSI performance. This study addresses this gap by systematically evaluating MIL feature extractors across three dimensions: pre-training dataset, backbone model, and pre-training method. Extensive experiments were conducted on two public WSI datasets (TCGA-NSCLC and Camelyon16) using four state-of-the-art (SOTA) MIL models. Our findings reveal that: 1) selecting a robust self-supervised learning (SSL) method has a greater impact on performance than relying solely on an in-domain pre-training dataset; 2) prioritizing Transformer-based backbones with deeper architectures over CNN-based models; and 3) using larger, more diverse pre-training datasets significantly enhances classification outcomes. We hope that these insights can provide practical guidance for optimizing WSI classification and explain the reasons behind the performance advantages of the current SOTA pathology foundation models. Furthermore, this work may inform the development of more effective pathology foundation models. Our code is publicly available at https://github.com/bryanwong17/MIL-Feature-Extractor-Selection
Reliably detecting when a deployed machine learning model is likely to fail on a given input is crucial for ensuring safe operation. In this work, we propose DECIDER (Debiasing Classifiers to Identify Errors Reliably), a novel approach that leverages priors from large language models (LLMs) and vision-language models (VLMs) to detect failures in image classification models. DECIDER utilizes LLMs to specify task-relevant core attributes and constructs a ``debiased'' version of the classifier by aligning its visual features to these core attributes using a VLM, and detects potential failure by measuring disagreement between the original and debiased models. In addition to proactively identifying samples on which the model would fail, DECIDER also provides human-interpretable explanations for failure through a novel attribute-ablation strategy. Through extensive experiments across diverse benchmarks spanning subpopulation shifts (spurious correlations, class imbalance) and covariate shifts (synthetic corruptions, domain shifts), DECIDER consistently achieves state-of-the-art failure detection performance, significantly outperforming baselines in terms of the overall Matthews correlation coefficient as well as failure and success recall. Our codes can be accessed at~\url{https://github.com/kowshikthopalli/DECIDER/}
Existing explanation methods for image classification struggle to provide faithful and plausible explanations. This paper addresses this issue by proposing a post-hoc natural language explanation method that can be applied to any CNN-based classifier without altering its training process or affecting predictive performance. By analysing influential neurons and the corresponding activation maps, the method generates a faithful description of the classifier's decision process in the form of a structured meaning representation, which is then converted into text by a language model. Through this pipeline approach, the generated explanations are grounded in the neural network architecture, providing accurate insight into the classification process while remaining accessible to non-experts. Experimental results show that the NLEs constructed by our method are significantly more plausible and faithful. In particular, user interventions in the neural network structure (masking of neurons) are three times more effective than the baselines.
The Vision Transformer (ViT) leverages the Transformer's encoder to capture global information by dividing images into patches and achieves superior performance across various computer vision tasks. However, the self-attention mechanism of ViT captures the global context from the outset, overlooking the inherent relationships between neighboring pixels in images or videos. Transformers mainly focus on global information while ignoring the fine-grained local details. Consequently, ViT lacks inductive bias during image or video dataset training. In contrast, convolutional neural networks (CNNs), with their reliance on local filters, possess an inherent inductive bias, making them more efficient and quicker to converge than ViT with less data. In this paper, we present a lightweight Depth-Wise Convolution module as a shortcut in ViT models, bypassing entire Transformer blocks to ensure the models capture both local and global information with minimal overhead. Additionally, we introduce two architecture variants, allowing the Depth-Wise Convolution modules to be applied to multiple Transformer blocks for parameter savings, and incorporating independent parallel Depth-Wise Convolution modules with different kernels to enhance the acquisition of local information. The proposed approach significantly boosts the performance of ViT models on image classification, object detection, and instance segmentation by a large margin, especially on small datasets, as evaluated on CIFAR-10, CIFAR-100, Tiny-ImageNet and ImageNet for image classification, and COCO for object detection and instance segmentation. The source code can be accessed at https://github.com/ZTX-100/Efficient_ViT_with_DW.
Contrastive Language-Image Pre-training (CLIP), which excels at abstracting open-world representations across domains and modalities, has become a foundation for a variety of vision and multimodal tasks. However, recent studies reveal that CLIP has severe visual shortcomings, such as which can hardly distinguish orientation, quantity, color, structure, etc. These visual shortcomings also limit the perception capabilities of multimodal large language models (MLLMs) built on CLIP. The main reason could be that the image-text pairs used to train CLIP are inherently biased, due to the lack of the distinctiveness of the text and the diversity of images. In this work, we present a simple post-training approach for CLIP models, which largely overcomes its visual shortcomings via a self-supervised diffusion process. We introduce DIVA, which uses the DIffusion model as a Visual Assistant for CLIP. Specifically, DIVA leverages generative feedback from text-to-image diffusion models to optimize CLIP representations, with only images (without corresponding text). We demonstrate that DIVA improves CLIP's performance on the challenging MMVP-VLM benchmark which assesses fine-grained visual abilities to a large extent (e.g., 3-7%), and enhances the performance of MLLMs and vision models on multimodal understanding and segmentation tasks. Extensive evaluation on 29 image classification and retrieval benchmarks confirms that our framework preserves CLIP's strong zero-shot capabilities. The code is available at https://github.com/baaivision/DIVA.
Vision transformers have significantly advanced the field of computer vision, offering robust modeling capabilities and global receptive field. However, their high computational demands limit their applicability in processing long sequences. To tackle this issue, State Space Models (SSMs) have gained prominence in vision tasks as they offer linear computational complexity. Recently, State Space Duality (SSD), an improved variant of SSMs, was introduced in Mamba2 to enhance model performance and efficiency. However, the inherent causal nature of SSD/SSMs restricts their applications in non-causal vision tasks. To address this limitation, we introduce Visual State Space Duality (VSSD) model, which has a non-causal format of SSD. Specifically, we propose to discard the magnitude of interactions between the hidden state and tokens while preserving their relative weights, which relieves the dependencies of token contribution on previous tokens. Together with the involvement of multi-scan strategies, we show that the scanning results can be integrated to achieve non-causality, which not only improves the performance of SSD in vision tasks but also enhances its efficiency. We conduct extensive experiments on various benchmarks including image classification, detection, and segmentation, where VSSD surpasses existing state-of-the-art SSM-based models. Code and weights are available at \url{https://github.com/YuHengsss/VSSD}.