论文检索

Sanjoy Chowdhury, Subrata Biswas, Sayan Nag, Tushar Nagarajan, Calvin Murdock, Ishwarya Ananthabhotla, Yijun Qian, Vamsi Krishna Ithapu, Dinesh Manocha, Ruohan Gao

Modern perception models, particularly those designed for multisensory egocentric tasks, have achieved remarkable performance but often come with substantial computational costs. These high demands pose challenges for real-world deployment, especially in resource-constrained environments. In this paper, we introduce EGOADAPT, a framework that adaptively performs cross-modal distillation and policy learning to enable efficient inference across different egocentric perception tasks, including egocentric action recognition, active speaker localization, and behavior anticipation. Our proposed policy module is adaptable to task-specific action spaces, making it broadly applicable. Experimental results on three challenging egocentric datasets - EPIC-Kitchens, EasyCom, and Aria Everyday Activities - demonstrate that our method significantly enhances efficiency, reducing GMACs by up to 89.09%, parameters up to 82.02%, and energy up to 9.6x, while still on-par and in many cases outperforming the performance of corresponding state-of-the-art models.

Seongmin Park, Hyungmin Kim, Sangwoo Kim, Wonseok Jeon, Juyoung Yang, Byeongwook Jeon, Yoonseon Oh, Jungwook Choi

Deep neural network (DNN)-based policy models, such as vision-language-action (VLA) models, excel at automating complex decision-making from multi-modal inputs. However, scaling these models greatly increases computational overhead, complicating deployment in resource-constrained settings like robot manipulation and autonomous driving. To address this, we propose Saliency-Aware Quantized Imitation Learning (\method), which combines quantization-aware training with a selective loss-weighting strategy for mission-critical states. By identifying these states via saliency scores and emphasizing them in the training loss, \method preserves decision fidelity under low-bit precision. We validate \method's generalization capability across extensive simulation benchmarks with environment variations, real-world tasks, and cross-domain tasks (self-driving, physics simulation), consistently recovering full-precision performance. Notably, a 4-bit weight-quantized VLA model for robotic manipulation achieves up to 2.5xspeedup and 2.5xenergy savings on an edge GPU with minimal accuracy loss. These results underline \method's potential for efficiently deploying large IL-based policy models on resource-limited devices.

Zhenjun Yu, Wenqiang Xu, Pengfei Xie, Yutong Li, Brian W. Anthony, Zhuorui Zhang, Cewu Lu

We present ViTaM-D, a novel visual-tactile framework for reconstructing dynamic hand-object interaction with distributed tactile sensing to enhance contact modeling. Existing methods, relying solely on visual inputs, often fail to capture occluded interactions and object deformation. To address this, we introduce DF-Field, a distributed force-aware contact representation leveraging kinetic and potential energy in hand-object interactions. ViTaM-D first reconstructs interactions using a visual network with contact constraint, then refines contact details through force-aware optimization, improving object deformation modeling. To evaluate deformable object reconstruction, we introduce the HOT dataset, featuring 600 hand-object interaction sequences in a high-precision simulation environment. Experiments on DexYCB and HOT datasets show that ViTaM-D outperforms state-of-the-art methods in reconstruction accuracy for both rigid and deformable objects. DF-Field also proves more effective in refining hand poses and enhancing contact modeling than previous refinement methods. The code, models, and datasets are available at https://sites.google.com/view/vitam-d/.

Chen Zhu, Wangbo Zhao, Huiwen Zhang, Yuhao Zhou, Weidong Tang, Shuo Wang, Zhihang Yuan, Yuzhang Shang, Xiaojiang Peng, Kai Wang 等

Vision Transformers (ViTs) have emerged as a foundational model in computer vision, excelling in generalization and adaptation to downstream tasks. However, deploying ViTs to support diverse resource constraints typically requires retraining multiple, size-specific ViTs, which is both time-consuming and energy-intensive. To address this issue, we propose an efficient ViT adaptation framework that enables a single adaptation process to generate multiple models of varying sizes for deployment on platforms with various resource constraints. Our approach comprises two stages. In the first stage, we enhance a pre-trained ViT with a nested elastic architecture that enables structural flexibility across MLP expansion ratio, number of attention heads, embedding dimension, and network depth. To preserve pre-trained knowledge and ensure stable adaptation, we adopt a curriculum-based training strategy that progressively increases elasticity. In the second stage, we design a lightweight router to select submodels according to computational budgets and downstream task demands. Initialized with Pareto-optimal configurations derived via a customized NSGA-II algorithm, the router is then jointly optimized with the backbone. Extensive experiments on multiple benchmarks demonstrate the effectiveness and versatility of EA-ViT. The code is available at https://github.com/zcxcf/EA-ViT.

Wenlun Zhang, Yunshan Zhong, Shimpei Ando, Kentaro Yoshioka

The Segment Anything Model (SAM) has demonstrated strong versatility across various visual tasks. However, its large storage requirements and high computational cost pose challenges for practical deployment. Post-training quantization (PTQ) has emerged as an effective strategy for efficient deployment, but we identify two key challenges in SAM that hinder the effectiveness of existing PTQ methods: the heavy-tailed and skewed distribution of post-GELU activations, and significant inter-channel variation in linear projection activations. To address these challenges, we propose AHCPTQ, an accurate and hardware-efficient PTQ method for SAM. AHCPTQ introduces hardware-compatible Hybrid Log-Uniform Quantization (HLUQ) to manage post-GELU activations, employing log2 quantization for dense small values and uniform quantization for sparse large values to enhance quantization resolution. Additionally, AHCPTQ incorporates Channel-Aware Grouping (CAG) to mitigate inter-channel variation by progressively clustering activation channels with similar distributions, enabling them to share quantization parameters and improving hardware efficiency. The combination of HLUQ and CAG not only enhances quantization effectiveness but also ensures compatibility with efficient hardware execution. For instance, under the W4A4 configuration on the SAM-L model, AHCPTQ achieves 36.6% mAP on instance segmentation with the DINO detector, while achieving a 7.89x speedup and 8.64x energy efficiency over its floating-point counterpart in FPGA implementation.

Tomoya Sawada

Balancing energy efficiency and occupant comfort in building HVAC systems is a critical challenge. While generative AI shows promise, its application has been hindered by a reliance on simulations and the inherent instability of its numerical predictions. This paper presents ''Office-in-the-Loop,'' a cyber-physical system leveraging generative AI in a real-world office. Our real-world experiments resolve the energy-comfort trade-off, achieving up to 47.92% energy savings with a 26.36% comfort improvement. We introduce a novel prompting technique, ''Data-Driven Reasoning,'' which compels the AI to justify its predictions with data. This simple addition improves prediction accuracy within ±0.5°C from 50% to 92.31%, paving the way for reliable, AI-driven building automation.

Ashan Perera, Md Eimran Hossain Eimon, Juan Merlos, Velibor Adzic, Hari Kalva, Borko Furht

As numerous edge devices start implementing intelligent components, the challenges of energy consumption, bandwidth efficiency, and privacy gain significance. One proposed solution relies on the paradigm of split inference, which optimizes the delegation of the computational load between edge and remote devices. We developed and implemented the standard-compliant split inference system with an encoder and decoder capable of real-time streaming and processing. Our system outperforms state-of-the-art video compression implementations by an average of 83% bitrate reduction, while preserving privacy. We demonstrate the system's real-time performance on consumer devices, with interactive visualizations of object detection and segmentation, incorporating real-time metrics. Demo video: https://youtu.be/bmCbUo_ZWWU

Keyue Shi, Qianqian Shen, Zhaoming Ye, Liangjun Jiang, Jiajun Bu, Haishuai Wang

Osteoporosis has emerged as a significant global public health challenge, affecting more than 200 million people worldwide. Precise screening through bone mineral density (BMD) assessment is essential for timely intervention. Although dual-energy X-ray absorptiometry (DXA) serves as the gold standard for BMD evaluation, its widespread application is hampered by high costs and limited accessibility. In recent years, deep learning has demonstrated great potential in osteoporosis classification and predicting BMD from X-ray and CT scans. However, existing public datasets have notable limitations, particularly the absence of paired multimodal imaging datasets for cross-modal modeling and the lack of accurate BMD annotations required for osteoporosis-specific research and clinical model optimization. To address these issues, we introduce the Lumbar Multimodal Osteoporosis Screening dataset (LUMOS), the first multimodal dataset specifically designed for lumbar osteoporosis screening. LUMOS integrates clinical data from 803 patients, including 1,620 anteroposterior/lateral lumbar X-rays with BMD values and T-scores, comprehensive demographic information, and 280 lumbar CT scans. The advent of LUMOS is expected to propel forward research on automated osteoporosis classification, BMD prediction, and other related tasks. Its standardized and multimodal nature fills critical gaps in lumbar osteoporosis data, providing high-quality data support for the development and validation of medical AI algorithms in the early detection of osteoporosis. The dataset is available at https://keyueshi.github.io/LUMOS/.

Haizhou Wang, Guobing Zou, Fei Xu 0009, Yangguang Cui, Tongquan Wei

Federated learning (FL), an emerging data-secure distributed training paradigm, unites massive isolated Internet of Things (IoT) device nodes to collaboratively train a global neural network (NN) model without the exposure of their local multimedia data. However, constrained by the synchronous NN model integration nature of FL, there is a training latency inconsistency among heterogeneous devices, which significantly deteriorates FL training efficiency. Meanwhile, frequent local NN training and transmission impose high energy consumption pressure on users. To tackle these issues, this paper proposes a premium multi-width NN-assisted hierarchical FL (HFL) framework in heterogeneous cloud-edge-device computing to achieve remarkable training speedup and energy conservation. Specifically, a heterogeneity-aware NN width coefficient determination algorithm, which flexibly assigns a subnet with a suitable width to each user device based on its computing ability, is first applied to shorten the HFL training latency. Subsequently, to integrate subnets with different width topologies, we design a width-aware adaptive NN model integration approach to effectively ensure the accuracy of the integrated global NN model. Finally, a latency-aware energy saving strategy is introduced to reduce energy consumption. Experimental results demonstrate that our proposed framework outperforms state-of-the-art benchmarks, and attains up to 42.42% enhancement in accuracy, 81.5% reduction in training latency, and 40.9% optimization in energy cost.

Kewei Zhao, Xiaowei Hu 0001, Qinya Li

Unknown object detection aims to build detectors capable of identifying out-of-distribution objects, a critical need for applications like autonomous driving and traffic monitoring. However, limited device resources restrict existing methods from achieving accurate detection on the device side. Addressing this gap, this paper introduces a device-cloud collaborative framework named DCCUOD that enhances device model performance through efficient cloud collaboration. Our framework employs an energy-based sampling function on devices to target samples with unknown objects, coupled with a collaborative pseudo-labeling strategy to generate accurate pseudo-labels. Additionally, a two-stage training paradigm enables continuous improvements of device models on both known and unknown objects. Our study is the first to explore device-cloud collaborative learning for UOD tasks. Experimental results show that the device model is three times smaller and seven times faster than cloud models, with minimal performance trade-offs.

Yang Zhao 0002, Shusheng Li, Xueshang Feng

As the development of lightweight deep learning algorithms, various deep neural network (DNN) models have been proposed for the remote sensing scene classification (RSSC) application. However, it is still challenging for these RSSC models to achieve optimal performance among model accuracy, inference latency, and energy consumption on resource-constrained edge devices. In this paper, we propose a lightweight RSSC framework, which includes a distilled global filter network (GFNet) model and an early-exit mechanism designed for edge devices to achieve state-of-the-art performance. Specifically, we first apply frequency domain distillation on the GFNet model to reduce model size. Then we design a dynamic early-exit model tailored for DNN models on edge devices to further improve model inference efficiency. We evaluate our E3C model on three edge devices across four datasets. Extensive experimental results show that it achieves an average of 1.3x speedup on model inference and over 40% improvement on energy efficiency, while maintaining high classification accuracy.

Pengyu Zeng, Jun Yin, Haoyuan Sun, Yuqin Dai, Maowei Jiang, Miao Zhang 0010, Shuai Lu

Residential design is a complex and open-ended problem that requires designers to integrate diverse types of input information while adhering to stringent energy consumption standards. However, most current research in this field focuses on generating floor plans from a limited set of input types, often neglecting to incorporate energy-related physical constraints. Existing approaches are limited by: (1) the lack of multimodal datasets in this domain, (2) the absence of comprehensive residential energy consumption data, and (3) the challenges associated with effectively integrating multiple input types into a unified model. To address these challenges, we propose MRED-14, the first large-scale Multimodal Residential Energy Dataset, comprising 14 input types, including energy consumption values, vector drawings, and textual descriptions, paired with 41,280 high-quality residential floor plans that have been scored and annotated by human experts. Based on this dataset, we introduce the LER-net model, which can flexibly adapt to various input types and generate low-energy residential floor plans. Experimental results demonstrate that LER-net outperforms existing models, achieving state-of-the-art performance under the same input conditions. In addition, the energy consumption of the generated floor plans is reduced by 5.1% compared to the actual residential designs. Further expert evaluations confirm the LER-net model's feasibility for use in residential design.

Jiangrong Shen, Yulin Xie, Qi Xu 0008, Gang Pan 0001, Huajin Tang, Badong Chen

Multimodal spiking neural networks (SNNs) hold significant potential for energy-efficient sensory processing but face critical challenges in modality imbalance and temporal misalignment. Current approaches suffer from uncoordinated convergence speeds across modalities and static fusion mechanisms that ignore time-varying cross-modal interactions. We propose the temporal attention-guided adaptive fusion framework for multimodal SNNs with two synergistic innovations: 1) The Temporal Attention-guided Adaptive Fusion (TAAF) module that dynamically assigns importance scores to fused spiking features at each timestep, enabling hierarchical integration of temporally heterogeneous spike-based features; 2) The temporal adaptive balanced fusion loss that modulates learning rates per modality based on the above attention scores, preventing dominant modalities from monopolizing optimization. The proposed framework implements adaptive fusion, especially in the temporal dimension, and alleviates the modality imbalance during multimodal learning, mimicking cortical multisensory integration principles. Evaluations on CREMA-D, AVE, and EAD datasets demonstrate state-of-the-art performance (77.55%, 70.65% and 98.65% accuracy, respectively) with energy efficiency. The system resolves temporal misalignment through learnable time-warping operations and faster modality convergence coordination than baseline SNNs. This work establishes a new paradigm for temporally coherent multimodal learning in neuromorphic systems, bridging the gap between biological sensory processing and efficient machine intelligence.mfp

Yang Li, Xinyi Zeng, Zhe Xue, Pinxian Zeng, Zikai Zhang, Yan Wang 0015

As the third generation of neural networks, spiking neural networks (SNNs) have recently gained widespread attention for their biological plausibility, energy efficiency, and effectiveness in processing neuromorphic datasets. To better emulate biological neurons, various models such as Integrate-and-Fire (IF) and Leaky Integrate-and-Fire (LIF) have been widely adopted in SNNs. However, these neuron models overlook the refractory period, a fundamental characteristic of biological neurons. Research on excitable neurons reveal that after firing, neurons enter a refractory period during which they are temporarily unresponsive to subsequent stimuli. This mechanism is critical for preventing over-excitation and mitigating interference from aberrant signals. Therefore, we propose a simple yet effective method to incorporate the refractory period into spiking LIF neurons through spike-triggered threshold dynamics, termed RPLIF. Our method ensures that each spike accurately encodes neural information, effectively preventing neuron over-excitation under continuous inputs and interference from anomalous inputs. Incorporating the refractory period into LIF neurons is seamless and computationally efficient, enhancing robustness and efficiency while yielding better performance with negligible overhead. To the best of our knowledge, RPLIF achieves state-of-the-art performance on Cifar10-DVS(82.40%) and N-Caltech101(83.35%) with fewer timesteps and demonstrates superior performance on DVS128 Gesture(97.22%) at low latency.

Meng Yu, Kun Zhan

Diffusion models exhibit impressive generative capabilities but are significantly impacted by exposure bias. In this paper, we make a key observation: the energy of predicted noisy samples in the reverse process continuously declines compared to perturbed samples in the forward process. Building on this, we identify two important findings: 1) The reduction in energy follows distinct patterns in the low-frequency and high-frequency subbands; 2) The subband energy of reverse-process reconstructed samples is consistently lower than that of forward-process ones, and both are lower than the original data samples. Based on the first finding, we introduce a dynamic frequency regulation mechanism utilizing wavelet transforms, which separately adjusts the low- and high-frequency subbands. Leveraging the second insight, we derive the rigorous mathematical form of exposure bias. It is worth noting that, our method is training-free and plug-and-play, significantly improving the generative quality of various diffusion models and frameworks with negligible computational cost. The source code is available at https://github.com/kunzhan/wpp.

Ruicheng Zhang, Haowei Guo, Kanghui Tian, Jun Zhou, Mingliang Yan, Zeyu Zhang 0006, Shen Zhao

Unified Medical Image Segmentation (UMIS) is critical for comprehensive anatomical assessment but faces challenges due to multi-scale structural heterogeneity. Conventional pixel-based approaches, lacking object-level anatomical insight and inter-organ relational modeling, struggle with morphological complexity and feature conflicts, limiting their efficacy in UMIS. We propose Mamba Snake, a novel deep snake framework enhanced by state space modeling for UMIS. Mamba Snake frames multi-contour evolution as a hierarchical state space atlas, effectively modeling macroscopic inter-organ topological relationships and microscopic contour refinements. We introduce a snake-specific vision state space module, the Mamba Evolution Block (MEB), which leverages effective spatiotemporal information aggregation for adaptive refinement of complex morphologies. Energy map shape priors further ensures robust long-range contour evolution in heterogeneous data. Additionally, a dual-classification synergy mechanism is incorporated to concurrently optimize detection and segmentation, mitigating under-segmentation of microstructures in UMIS. Extensive evaluations across five clinical datasets reveal Mamba Snake's superior performance.

Shuning Sun, Yu Zhang 0296, Chen Wu 0006, Dianjie Lu, Guijuan Zhang, Yang Wen, Zhuoran Zheng

Video imaging is often affected by complex degradations such as blur, noise, and compression artifacts. Traditional restoration methods follow a ''single-task single-model'' paradigm, resulting in poor generalization and high computational cost, limiting their applicability in real-world scenarios with diverse degradation types. We propose UniFlowRestore, a general video restoration framework that models restoration as a time-continuous evolution under a prompt-guided and physics-informed vector field. A physics-aware backbone PhysicsUNet encodes degradation priors as potential energy, while PromptGenerator produces task-relevant prompts as momentum. These components define a Hamiltonian system whose vector field integrates inertial dynamics, decaying physical gradients, and prompt-based guidance. The system is optimized via a fixed-step ODE solver to achieve efficient and unified restoration across tasks. Experiments show that UniFlowRestore delivers state-of-the-art performance with strong generalization and efficiency. Quantitative results demonstrate that UniFlowRestore achieves state-of-the-art performance, attaining the highest PSNR (33.89 dB) and SSIM (0.97) on the video denoising task, while maintaining top or second-best scores across all evaluated tasks.

Xuan Zhang, Sin Chee Chin, Jing-Hao Xue, Xiaochen Yang, Wenming Yang

Long-tailed out-of-distribution learning aims to reduce performance bias in long-tailed in-distribution (ID) data while rejecting out-of-distribution (OOD) samples, which are often mistaken for under-represented tail classes. To achieve OOD detection, existing methods incorporate an outlier exposure (OE) term into the long-tailed recognition (LTR) loss. However, as we prove in this paper, the OE term induces a gradient conflict with the ID objectives, especially for tail classes, thereby contradicting the core motivation of LTR. To avoid the ID-OOD dilemma, we propose Dynamic Ambiguity-aware Recalibration for Logits (DARL), an ambiguity-guided long-tailed OOD learning approach, grounded on two theoretical insights. First, we show that the mixed ID data can mitigate the conflict in OE training and exhibits higher intrinsic ambiguity than the original ID data, thus able to serve as a surrogate for real OOD data. Second, we introduce an ambiguity-aware logit adjustment that can dynamically calibrate the class margins using energy-based ambiguity metrics, effectively reducing early-stage bias while avoiding late-stage overfitting. Extensive experiments show that DARL achieves the overall state-of-the-art performance of long-tailed OOD learning. Moreover, compared with the OE methods, DARL trains solely on the ID data, which can reduce the data requirements by 80%. The code is available in https://github.com/XuanZhang-A/DARL.

Xinyu Zhang 0021, Lingling Zhang 0005, Yanrui Wu, Muye Huang, Jun Liu 0002

Abstract visual reasoning, exemplified by Raven's Progressive Matrices (RPM), remains a significant challenge in artificial intelligence. A critical difficulty lies in disentangling abstract relational rules from image-specific features, as these rules operate independently of visual appearances. To address this challenge, we propose the Cognitive Predictive Coding Network (CPCN), inspired by predictive coding theory from cognitive science. CPCN features a three-component architecture: a Relation Disentangler that separates abstract rules from image-specific features through prediction error minimization; Stacked Free Energy Minimizers (FEMs) that leverage energy minimization principles to progressively reduce uncertainty during hierarchical abstraction; and a classifier for solution identification. Unlike previous approaches, our model employs mutual information constraints to explicitly separate relation-relevant and relation-irrelevant features, enabling more robust pattern recognition. Our novel FEMs provide a principled approach to uncertainty reduction through iterative refinement of pattern understanding. Experiments demonstrate PAH's superior performance across multiple benchmarks (such as 98.9% on RAVEN-Fair) with state-of-the-art 59.7% average accuracy across all PGM subtasks.

Yang Ren 0001, Hai Jiang 0006, Wei Li 0075, Menglong Yang, Heng Zhang 0042, Zehua Sheng, Qingsheng Ye, Shuaicheng Liu

Image downscaling is critical for efficient storage and transmission of high-resolution (HR) images. Existing learning-based methods focus on performing downscaling within the sRGB domain, which typically suffers from blurred details and unexpected artifacts. RAW images, with their unprocessed photonic information, offer greater flexibility but lack specialized downscaling frameworks. In this paper, we propose a wavelet-based recurrent reconstruction framework that leverages the information lossless attribute of wavelet transformation to fulfill the arbitrary-scale RAW image downscaling in a coarse-to-fine manner, in which the Low-Frequency Arbitrary-Scale Downscaling Module (LASDM) and the High-Frequency Prediction Module (HFPM) are proposed to preserve structural and textural integrity of the reconstructed low-resolution (LR) RAW images, alongside an energy-maximization loss to align high-frequency energy between HR and LR domain. Furthermore, we introduce the Realistic Non-Integer RAW Downscaling (Real-NIRD) dataset, featuring a non-integer downscaling factor of 1.3×, and incorporate it with publicly available datasets with integer factors (2×, 3×, 4×) for comprehensive benchmarking arbitrary-scale image downscaling purposes. Extensive experiments demonstrate that our method outperforms existing state-of-the-art competitors both quantitatively and visually. The code and dataset will be released at https://github.com/RenYangSCU/ASRD.