Channel pruning is a promising method for accelerating and compressing convolutional neural networks. However, current pruning algorithms still remain unsolved problems that how to assign layer-wise pruning ratios properly and discard the least important channels with a convincing criterion. In this paper, we present a novel channel pruning approach via information theory and interpretability of neural networks. Specifically, we regard information entropy as the expected amount of information for convolutional layers. In addition, if we suppose a matrix as a system of linear equations, a higher-rank matrix represents there exist more solutions to it, which indicates more uncertainty. From the point of view of information theory, the rank can also describe the amount of information. In a neural network, considering the rank and entropy as two information indicators of convolutional layers, we propose a fusion function to reach a compromise of them, where the fusion results are defined as ``information concentration''. When pre-defining layer-wise pruning ratios, we employ the information concentration as a reference instead of heuristic and engineering tuning to provide a more interpretable solution. Moreover, we leverage Shapley values, which are a potent tool in the interpretability of neural networks, to evaluate the channel contributions and discard the least important channels for model compression while maintaining its performance. Extensive experiments demonstrate the effectiveness and promising performance of our method. For example, our method improves the accuracy by 0.21% when reducing 45.5% FLOPs and removing 40.3% parameters for ResNet-56 on CIFAR-10. Moreover, our method obtains loss in Top-1/Top-5 accuracies of 0.43%/0.11% by reducing 41.6% FLOPs and removing 35.0% parameters for ResNet-50 on ImageNet.
Deep neural networks typically impose significant computational loads and memory consumption. Moreover, the large parameters pose constraints on deploying the model on edge devices such as embedded systems. Tensor decomposition offers a clear advantage in compressing large-scale weight tensors. Nevertheless, direct utilization of low-rank decomposition typically leads to significant accuracy loss. This paper proposes a model compression method that integrates Variational Bayesian Matrix Factorization (VBMF) with orthogonal regularization. Initially, the model undergoes over-parameterization and training, with orthogonal regularization applied to enhance its likelihood of achieving the accuracy of the original model. Secondly, VBMF is employed to estimate the rank of the weight tensor at each layer. Our framework is sufficiently general to apply to other convolutional neural networks and easily adaptable to incorporate other tensor decomposition methods. Experimental results show that for both high and low compression ratios, our compression model exhibits advanced performance.
With the popularity of video applications, the security of video content has emerged as a pressing issue that demands urgent attention. Most video content protection methods mainly rely on encryption technology, which needs to be manually designed or implemented in an experience-based manner. To address this problem, we propose an automatic encryption technique for video content protection based on implicit neural representation. We design a key-controllable module, which serves as a key for encryption and decryption. NeR-VCP first pre-distributes the key-controllable module trained by the sender to the recipients, and then uses Implicit Neural Representation (INR) with a (pre-distributed) key-controllable module to encrypt plain video as an implicit neural network, and the legal recipients uses a pre-distributed key-controllable module to decrypt this cipher neural network (the corresponding implicit neural network). Under the guidance of the key-controllable design, our method can improve the security of video content and provide a novel video encryption scheme. Moreover, using model compression techniques, this method can achieve video content protection while effectively mitigating the amount of encrypted data transferred. We experimentally find that it has superior performance in terms of visual representation, imperceptibility to illegal users, and security from a cryptographic viewpoint.
Variational Autoencoders (VAEs), as a form of deep generative model, have been widely used in recent years, and shown great great peformance in a number of different domains, including image generation and anomaly detection, etc.. This paper aims to explore neural network model compression method based on VAE. The experiment uses different neural network models for MNIST recognition as compression targets, including Feedforward Neural Network (FNN), Convolutional Neural Network (CNN), Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM). These models are the most basic models in deep learning, and other more complex and advanced models are based on them or inherit their features and evolve. In the experiment, the first step is to train the models mentioned above, each trained model will have different accuracy and number of total parameters. And then the variants of parameters for each model are processed as training data in VAEs separately, and the trained VAEs are tested by the true model parameters. The experimental results show that using the latent space as a representation of the model compression can improve the compression rate compared to some traditional methods such as pruning and quantization, meanwhile the accuracy is not greatly affected using the model parameters reconstructed based on the latent space. In the future, a variety of different large-scale deep learning models will be used more widely, so exploring different ways to save time and space on saving or transferring models will become necessary, and the use of VAE in this paper can provide a basis for these further explorations.
Model merging offers an effective strategy to combine the strengths of multiple finetuned models into a unified model that preserves the specialized capabilities of each. Existing methods merge models in a global manner, performing arithmetic operations across all model parameters. However, such global merging often leads to task interference, degrading the performance of the merged model. In this work, we introduce Localize-and-Stitch, a novel approach that merges models in a localized way. Our algorithm works in two steps: i) Localization: identify tiny ($1\%$ of the total parameters) localized regions in the finetuned models containing essential skills for the downstream tasks, and ii) Stitching: reintegrate only these essential regions back into the pretrained model for task synergy. We demonstrate that our approach effectively locates sparse regions responsible for finetuned performance, and the localized regions could be treated as compact and interpretable representations of the finetuned models (tasks). Empirically, we evaluate our method on various vision and language benchmarks, showing that it outperforms existing model merging methods under different data availability scenarios. Beyond strong empirical performance, our algorithm also facilitates model compression and preserves pretrained knowledge, enabling flexible and continual skill composition from multiple finetuned models with minimal storage and computational overhead. Our code is available at https://github.com/uiuctml/Localize-and-Stitch.
Determining the optimal size of a neural network is critical, as it directly impacts runtime performance and memory usage. Pruning is a well-established model compression technique that reduces the size of neural networks while mathematically guaranteeing accuracy preservation. However, many recent pruning methods overlook the global contributions of individual model components, making it difficult to ensure that a pruned model meets the desired dataset and performance requirements. To address these challenges, we developed a new pruning algorithm, MPruner, that leverages mutual information through vector similarity. MPruner utilizes layer clustering with the Centered Kernel Alignment (CKA) similarity metric, allowing us to incorporate global information from the neural network for more precise and efficient layer-wise pruning. We evaluated MPruner across various architectures and configurations, demonstrating its versatility and providing practical guidelines. MPruner achieved up to a 50% reduction in parameters and memory usage for CNN and transformer-based models, with minimal to no loss in accuracy.
Federated Learning (FL) offers a promising approach for collaborative machine learning across distributed devices. However, its adoption is hindered by the complexity of building reliable communication architectures and the need for expertise in both machine learning and network programming. This paper presents a comprehensive solution that simplifies the orchestration of FL tasks while integrating intent-based automation. We develop a user-friendly web application supporting the federated averaging (FedAvg) algorithm, enabling users to configure parameters through an intuitive interface. The backend solution efficiently manages communication between the parameter server and edge nodes. We also implement model compression and scheduling algorithms to optimize FL performance. Furthermore, we explore intent-based automation in FL using a fine-tuned Language Model (LLM) trained on a tailored dataset, allowing users to conduct FL tasks using high-level prompts. We observe that the LLM-based automated solution achieves comparable test accuracy to the standard web-based solution while reducing transferred bytes by up to 64% and CPU time by up to 46% for FL tasks. Also, we leverage the neural architecture search (NAS) and hyperparameter optimization (HPO) using LLM to improve the performance. We observe that by using this approach test accuracy can be improved by 10-20% for the carried out FL tasks.
Drowsiness detection holds paramount importance in ensuring safety in workplaces or behind the wheel, enhancing productivity, and healthcare across diverse domains. Therefore accurate and real-time drowsiness detection plays a critical role in preventing accidents, enhancing safety, and ultimately saving lives across various sectors and scenarios. This comprehensive review explores the significance of drowsiness detection in various areas of application, transcending the conventional focus solely on driver drowsiness detection. We delve into the current methodologies, challenges, and technological advancements in drowsiness detection schemes, considering diverse contexts such as public transportation, healthcare, workplace safety, and beyond. By examining the multifaceted implications of drowsiness, this work contributes to a holistic understanding of its impact and the crucial role of accurate and real-time detection techniques in enhancing safety and performance. We identified weaknesses in current algorithms and limitations in existing research such as accurate and real-time detection, stable data transmission, and building bias-free systems. Our survey frames existing works and leads to practical recommendations like mitigating the bias issue by using synthetic data, overcoming the hardware limitations with model compression, and leveraging fusion to boost model performance. This is a pioneering work to survey the topic of drowsiness detection in such an entirely and not only focusing on one single aspect. We consider the topic of drowsiness detection as a dynamic and evolving field, presenting numerous opportunities for further exploration.
To solve ever more complex problems, Deep Neural Networks are scaled to billions of parameters, leading to huge computational costs. An effective approach to reduce computational requirements and increase efficiency is to prune unnecessary components of these often over-parameterized networks. Previous work has shown that attribution methods from the field of eXplainable AI serve as effective means to extract and prune the least relevant network components in a few-shot fashion. We extend the current state by proposing to explicitly optimize hyperparameters of attribution methods for the task of pruning, and further include transformer-based networks in our analysis. Our approach yields higher model compression rates of large transformer- and convolutional architectures (VGG, ResNet, ViT) compared to previous works, while still attaining high performance on ImageNet classification tasks. Here, our experiments indicate that transformers have a higher degree of over-parameterization compared to convolutional neural networks. Code is available at https://github.com/erfanhatefi/Pruning-by-eXplaining-in-PyTorch.
Since the invention of GPT2--1.5B in 2019, large language models (LLMs) have transitioned from specialized models to versatile foundation models. The LLMs exhibit impressive zero-shot ability, however, require fine-tuning on local datasets and significant resources for deployment. Traditional fine-tuning techniques with the first-order optimizers require substantial GPU memory that exceeds mainstream hardware capability. Therefore, memory-efficient methods are motivated to be investigated. Model compression techniques can reduce energy consumption, operational costs, and environmental impact so that to support sustainable artificial intelligence advancements. Additionally, large-scale foundation models have expanded to create images, audio, videos, and multi-modal contents, further emphasizing the need for efficient deployment. Therefore, we are motivated to present a comprehensive overview of the prevalent memory-efficient fine-tuning methods over the network edge. We also review the state-of-the-art literatures on model compression to provide a vision on deploying LLMs over the network edge.
Large Language Models (LLMs) have reshaped the landscape of artificial intelligence by demonstrating exceptional performance across various tasks. However, substantial computational requirements make their deployment challenging on devices with limited resources. Recently, compression methods using low-rank matrix techniques have shown promise, yet these often lead to degraded accuracy or introduce significant overhead in parameters and inference latency. This paper introduces \textbf{Mo}dular \textbf{De}composition (MoDeGPT), a novel structured compression framework that does not need recovery fine-tuning while resolving the above drawbacks. MoDeGPT partitions the Transformer block into modules comprised of matrix pairs and reduces the hidden dimensions via reconstructing the module-level outputs. MoDeGPT is developed based on a theoretical framework that utilizes three well-established matrix decomposition algorithms -- Nystr\"om approximation, CR decomposition, and SVD -- and applies them to our redefined transformer modules. Our comprehensive experiments show MoDeGPT, without backward propagation, matches or surpasses previous structured compression methods that rely on gradient information, and saves 98% of compute costs on compressing a 13B model. On \textsc{Llama}-2/3 and OPT models, MoDeGPT maintains 90-95% zero-shot performance with 25-30% compression rates. Moreover, the compression can be done on a single GPU within a few hours and increases the inference throughput by up to 46%.
With the wide application of diffusion model, the high cost of inference resources has became an important bottleneck for its universal application. Controllable generation, such as ControlNet, is one of the key research directions of diffusion model, and the research related to inference acceleration and model compression is more important. In order to solve this problem, this paper proposes a modal reparameterization method, RepControlNet, to realize the controllable generation of diffusion models without increasing computation. In the training process, RepControlNet uses the adapter to modulate the modal information into the feature space, copy the CNN and MLP learnable layers of the original diffusion model as the modal network, and initialize these weights based on the original weights and coefficients. The training process only optimizes the parameters of the modal network. In the inference process, the weights of the neutralization original diffusion model in the modal network are reparameterized, which can be compared with or even surpass the methods such as ControlNet, which use additional parameters and computational quantities, without increasing the number of parameters. We have carried out a large number of experiments on both SD1.5 and SDXL, and the experimental results show the effectiveness and efficiency of the proposed RepControlNet.
Large Language Models (LLMs) have revolutionized natural language processing tasks. However, their practical application is constrained by substantial memory and computational demands. Post-training quantization (PTQ) is considered an effective method to accelerate LLM inference. Despite its growing popularity in LLM model compression, PTQ deployment faces two major challenges. First, low-bit quantization leads to performance degradation. Second, restricted by the limited integer computing unit type on GPUs, quantized matrix operations with different precisions cannot be effectively accelerated. To address these issues, we introduce a novel arbitrary-bit quantization algorithm and inference framework, ABQ-LLM. It achieves superior performance across various quantization settings and enables efficient arbitrary-precision quantized inference on the GPU. ABQ-LLM introduces several key innovations: (1) a distribution correction method for transformer blocks to mitigate distribution differences caused by full quantization of weights and activations, improving performance at low bit-widths. (2) the bit balance strategy to counteract performance degradation from asymmetric distribution issues at very low bit-widths (e.g., 2-bit). (3) an innovative quantization acceleration framework that reconstructs the quantization matrix multiplication of arbitrary precision combinations based on BTC (Binary TensorCore) equivalents, gets rid of the limitations of INT4/INT8 computing units. ABQ-LLM can convert each component bit width gain into actual acceleration gain, maximizing performance under mixed precision(e.g., W6A6, W2A8). Based on W2*A8 quantization configuration on LLaMA-7B model, it achieved a WikiText2 perplexity of 7.59 (2.17$\downarrow $ vs 9.76 in AffineQuant). Compared to SmoothQuant, we realized 1.6$\times$ acceleration improvement and 2.7$\times$ memory compression gain.
Deep neural networks have consistently represented the state of the art in most computer vision problems. In these scenarios, larger and more complex models have demonstrated superior performance to smaller architectures, especially when trained with plenty of representative data. With the recent adoption of Vision Transformer (ViT) based architectures and advanced Convolutional Neural Networks (CNNs), the total number of parameters of leading backbone architectures increased from 62M parameters in 2012 with AlexNet to 7B parameters in 2024 with AIM-7B. Consequently, deploying such deep architectures faces challenges in environments with processing and runtime constraints, particularly in embedded systems. This paper covers the main model compression techniques applied for computer vision tasks, enabling modern models to be used in embedded systems. We present the characteristics of compression subareas, compare different approaches, and discuss how to choose the best technique and expected variations when analyzing it on various embedded devices. We also share codes to assist researchers and new practitioners in overcoming initial implementation challenges for each subarea and present trends for Model Compression. Case studies for compression models are available at \href{https://github.com/venturusbr/cv-model-compression}{https://github.com/venturusbr/cv-model-compression}.
Model compression and hardware acceleration are essential for the resource-efficient deployment of deep neural networks. Modern object detectors have highly interconnected convolutional layers with concatenations. In this work, we study how pruning can be applied to such architectures, exemplary for YOLOv7. We propose a method to handle concatenation layers, based on the connectivity graph of convolutional layers. By automating iterative sensitivity analysis, pruning, and subsequent model fine-tuning, we can significantly reduce model size both in terms of the number of parameters and FLOPs, while keeping comparable model accuracy. Finally, we deploy pruned models to FPGA and NVIDIA Jetson Xavier AGX. Pruned models demonstrate a 2x speedup for the convolutional layers in comparison to the unpruned counterparts and reach real-time capability with 14 FPS on FPGA. Our code is available at https://github.com/fzi-forschungszentrum-informatik/iterative-yolo-pruning.
Recent research on knowledge distillation has increasingly focused on logit distillation because of its simplicity, effectiveness, and versatility in model compression. In this paper, we introduce Refined Logit Distillation (RLD) to address the limitations of current logit distillation methods. Our approach is motivated by the observation that even high-performing teacher models can make incorrect predictions, creating a conflict between the standard distillation loss and the cross-entropy loss. This conflict can undermine the consistency of the student model's learning objectives. Previous attempts to use labels to empirically correct teacher predictions may undermine the class correlation. In contrast, our RLD employs labeling information to dynamically refine teacher logits. In this way, our method can effectively eliminate misleading information from the teacher while preserving crucial class correlations, thus enhancing the value and efficiency of distilled knowledge. Experimental results on CIFAR-100 and ImageNet demonstrate its superiority over existing methods. The code is provided at \text{https://github.com/zju-SWJ/RLD}.
Although convolutional neural networks have made outstanding achievements in visible light target detection, there are still many challenges in infrared small object detection because of the low signal-to-noise ratio, incomplete object structure, and a lack of reliable infrared small object dataset. To resolve limitations of the infrared small object dataset, a new dataset named InfraTiny was constructed, and more than 85% bounding box is less than 32x32 pixels (3218 images and a total of 20,893 bounding boxes). A multi-scale attention mechanism module (MSAM) and a Feature Fusion Augmentation Pyramid Module (FFAFPM) were proposed and deployed onto embedded devices. The MSAM enables the network to obtain scale perception information by acquiring different receptive fields, while the background noise information is suppressed to enhance feature extraction ability. The proposed FFAFPM can enrich semantic information, and enhance the fusion of shallow feature and deep feature, thus false positive results have been significantly reduced. By integrating the proposed methods into the YOLO model, which is named Infra-YOLO, infrared small object detection performance has been improved. Compared to yolov3, [email omitted] has been improved by 2.7%; and compared to yolov4, that by 2.5% on the InfraTiny dataset. The proposed Infra-YOLO was also transferred onto the embedded device in the unmanned aerial vehicle (UAV) for real application scenarios, where the channel pruning method is adopted to reduce FLOPs and to achieve a tradeoff between speed and accuracy. Even if the parameters of Infra-YOLO are reduced by 88% with the pruning method, a gain of 0.7% is still achieved on [email omitted] compared to yolov3, and a gain of 0.5% compared to yolov4. Experimental results show that the proposed MSAM and FFAFPM method can improve infrared small object detection performance compared with the previous benchmark method.
In the domain of multimedia and multimodal processing, the efficient handling of diverse data streams such as images, video, and sensor data is paramount. Model compression and multitask learning (MTL) are crucial in this field, offering the potential to address the resource-intensive demands of processing and interpreting multiple forms of media simultaneously. However, effectively compressing a multitask model presents significant challenges due to the complexities of balancing sparsity allocation and accuracy performance across multiple tasks. To tackle these challenges, we propose AdapMTL, an adaptive pruning framework for MTL models. AdapMTL leverages multiple learnable soft thresholds independently assigned to the shared backbone and the task-specific heads to capture the nuances in different components' sensitivity to pruning. During training, it co-optimizes the soft thresholds and MTL model weights to automatically determine the suitable sparsity level at each component to achieve both high task accuracy and high overall sparsity. It further incorporates an adaptive weighting mechanism that dynamically adjusts the importance of task-specific losses based on each task's robustness to pruning. We demonstrate the effectiveness of AdapMTL through comprehensive experiments on popular multitask datasets, namely NYU-v2 and Tiny-Taskonomy, with different architectures, showcasing superior performance compared to state-of-the-art pruning methods.
3D Gaussian splatting (3DGS) has recently emerged as an alternative representation that leverages a 3D Gaussian-based representation and introduces an approximated volumetric rendering, achieving very fast rendering speed and promising image quality. Furthermore, subsequent studies have successfully extended 3DGS to dynamic 3D scenes, demonstrating its wide range of applications. However, a significant drawback arises as 3DGS and its following methods entail a substantial number of Gaussians to maintain the high fidelity of the rendered images, which requires a large amount of memory and storage. To address this critical issue, we place a specific emphasis on two key objectives: reducing the number of Gaussian points without sacrificing performance and compressing the Gaussian attributes, such as view-dependent color and covariance. To this end, we propose a learnable mask strategy that significantly reduces the number of Gaussians while preserving high performance. In addition, we propose a compact but effective representation of view-dependent color by employing a grid-based neural field rather than relying on spherical harmonics. Finally, we learn codebooks to compactly represent the geometric and temporal attributes by residual vector quantization. With model compression techniques such as quantization and entropy coding, we consistently show over 25x reduced storage and enhanced rendering speed compared to 3DGS for static scenes, while maintaining the quality of the scene representation. For dynamic scenes, our approach achieves more than 12x storage efficiency and retains a high-quality reconstruction compared to the existing state-of-the-art methods. Our work provides a comprehensive framework for 3D scene representation, achieving high performance, fast training, compactness, and real-time rendering. Our project page is available at https://maincold2.github.io/c3dgs/.
Deep Neural networks (DNNs), extensively applied across diverse disciplines, are characterized by their integrated and monolithic architectures, setting them apart from conventional software systems. This architectural difference introduces particular challenges to maintenance tasks, such as model restructuring (e.g., model compression), re-adaptation (e.g., fitting new samples), and incremental development (e.g., continual knowledge accumulation). Prior research addresses these challenges by identifying task-critical neuron layers, and dividing neural networks into semantically-similar sequential modules. However, such layer-level approaches fail to precisely identify and manipulate neuron-level semantic components, restricting their applicability to finer-grained model maintenance tasks. In this work, we implement NeuSemSlice, a novel framework that introduces the semantic slicing technique to effectively identify critical neuron-level semantic components in DNN models for semantic-aware model maintenance tasks. Specifically, semantic slicing identifies, categorizes and merges critical neurons across different categories and layers according to their semantic similarity, enabling their flexibility and effectiveness in the subsequent tasks. For semantic-aware model maintenance tasks, we provide a series of novel strategies based on semantic slicing to enhance NeuSemSlice. They include semantic components (i.e., critical neurons) preservation for model restructuring, critical neuron tuning for model re-adaptation, and non-critical neuron training for model incremental development. A thorough evaluation has demonstrated that NeuSemSlice significantly outperforms baselines in all three tasks.