This paper mainly studies the use of millimeter wave (mmWave) radar for 3D human pose estimation. Although pioneering works have achieved remarkable success, the lack of stability of continuous data acquisition by mmW...
详细信息
Out-of-distribution (OOD) detection is crucial in many real-world applications. However, intelligent models are often trained solely on in-distribution (ID) data, leading to overconfidence when misclassifying OOD data...
详细信息
Current intelligent diagnostic systems often catastrophically forget old knowledge when learning new diseases only from the training dataset of the new diseases. Inspired by human learning of visual classes with the e...
详细信息
ISBN:
(数字)9798350386226
ISBN:
(纸本)9798350386233
Current intelligent diagnostic systems often catastrophically forget old knowledge when learning new diseases only from the training dataset of the new diseases. Inspired by human learning of visual classes with the effective help of language, we propose a continual learning framework based on a pre-trained visual-language model (VLM) without storing any image of previously learned diseases. In this framework, textual prior knowledge of each new disease can be obtained by utilizing the frozen VLM’s text encoder, and then used to guide the visual learning of the new disease. This framework innovatively utilizes the textual prior knowledge of all previously learned diseases as out-of-distribution (OOD) information to help differentiate currently being-learned diseases from others. Extensive empirical evaluations on both medical and natural image datasets confirm the superiority of the proposed method over existing state-of-the-art methods in continual learning of new visual classes. The source code is available at https://***/OpenMedIA/TexCIL.
Medical rehabilitation robots have shown strong potential for rehabilitation assessment and monitoring of the elderly. Accurate estimation of human posture based on video analysis can help the clinical application of ...
详细信息
This paper proposes an improved consensus algorithm based on PBFT(EBCR-PBFT). Firstly, The Modified Random Select(MRS) function is used to perform preliminary screening of network nodes, so as to solve the problem of ...
详细信息
Preserving details and avoiding high computational costs are the two main challenges for the High-Resolution Salient Object Detection (HRSOD) task. In this paper, we propose a two-stage HRSOD model from the perspectiv...
详细信息
Preserving details and avoiding high computational costs are the two main challenges for the High-Resolution Salient Object Detection (HRSOD) task. In this paper, we propose a two-stage HRSOD model from the perspective of evolution and succession, including an evolution stage with Low-resolution Location Model (LrLM) and a succession stage with High-resolution Refinement Model (HrRM). The evolution stage achieves detail-preserving salient objects localization on the low-resolution image through the evolution mechanisms on supervision and feature;the succession stage utilizes the shallow high-resolution features to complement and enhance the features inherited from the first stage in a lightweight manner and generate the final high-resolution saliency prediction. Besides, a new metric named Boundary-Detail-aware Mean Absolute Error (MAEBD) is designed to evaluate the ability to detect details in high-resolution scenes. Extensive experiments on five datasets demonstrate that our network achieves superior performance at real-time speed (49 FPS) compared to state-of-the-art methods. Our code is publicly available at: https://***/rmcong/ESNet_ICML24. Copyright 2024 by the author(s)
The difficulty in fabricating a multifaceted composite heterojunction system based on CdxZn1-xS limits the enhancement of photocatalytic *** the present scrutiny,novel ZnO/CdxZn1-xS/CdS com-posite heterojunctions are ...
详细信息
The difficulty in fabricating a multifaceted composite heterojunction system based on CdxZn1-xS limits the enhancement of photocatalytic *** the present scrutiny,novel ZnO/CdxZn1-xS/CdS com-posite heterojunctions are successfully prepared by the alkaline dissolution etching *** internal electric field at the interface of Ⅰ-type and Z-scheme heterojunction improved the effective charge *** ZC 8 sample exhibits excellent photocatalytic performance and the H2 production efficiency is 15.67 mmol g-1 h-1 with good stability up to 82.9％in 24-hour *** performance of CH4 and CO capacity in the CO2RR process is 3.47 μmol g-1 h-1 and 23.5 μmol g-1 h-1,*** photogener-ated accelerated charge transport is then examined in detail by in situ X-ray photoelectron spectroscopy(ISXPS)and density functional theory(DFT)*** work presents a new idea for the synthe-sis of CdxZni-xS solid-solution-based materials and provides a solid reference for the detailed mechanism regarding the electric field at the heterojunction interface.
Large language models (LLMs) have demonstrated remarkable performance across a wide range of tasks, largely due to their substantial model size. However, this also results in significant GPU memory demands during infe...
详细信息
Large language models (LLMs) have demonstrated remarkable performance across a wide range of tasks, largely due to their substantial model size. However, this also results in significant GPU memory demands during inference. To address these challenges on hardware with limited GPU memory, existing approaches employ offloading techniques that offload unused tensors to CPU memory, thereby reducing GPU memory usage. Since offloading involves data transfer between GPU and CPU, it introduces transfer overhead. To mitigate this, prior works typically overlap data transfer with GPU computation using a fixed pipelining strategy applied uniformly across all inference iterations, referred to as static offloading. However, static offloading policies fail to maximize inference throughput because they cannot adapt to the dynamically changing transfer overhead during the inference process, leading to increasing GPU idleness and reduced inference *** propose that offloading policies should be adaptive to the varying transfer overhead across inference iterations to maximize inference throughput. To this end, we design and implement an adaptive offloading-based inference system called TightLLM with two key innovations. First, its key-value (KV) distributor employs a trade-compute-for-transfer strategy to address growing transfer overhead by dynamically recomputing portions of the KV cache, effectively overlapping data transfer with computation and minimizing GPU idleness. Second, TightLLM’s weight loader slices model weights and distributes the loading process across multiple batches, amortizing the excessive weight loading overhead and significantly improving throughput. Evaluation across various combinations of GPU hardware and LLM models shows that TightLLM achieves 1.3 to 23 times higher throughput during the decoding phase and 1.2 to 22 times higher throughput in the prefill phase compared to state-of-the-art offloading systems. Due to the higher throughput in prefill
Recommender systems based on Matrix Factorization are widely used. However, they can easily suffer from the problem of overrecommendation of popular items, i.e., popularity bias. To mitigate popularity bias, current m...
Recommender systems based on Matrix Factorization are widely used. However, they can easily suffer from the problem of overrecommendation of popular items, i.e., popularity bias. To mitigate popularity bias, current methods often uniformly model interactions' popularity bias degree considering user activity and interacted item's popularity, and then force the recommenders to focus more on less biased interactions. However, users' popularity preference for candidate items also plays an important role in estimating popularity bias, which is ignored by current methods. Therefore, their uniform modeling of bias degree results in sub-optimal debiasing performance. To address this issue, our core idea is to estimate personalized bias degrees to perform user-specific debiasing. We first derive a predefined bias degree obtained by items' popularity, then scale it considering users' candidate items' popularity. Extensive experiments conducted on two classic MF-based recommenders and three real-world datasets demonstrate that our approach outperforms state-of-the-art methods for popularity debiasing.
Weakly-supervised action segmentation is a task of learning to partition a long video into several action segments, where training videos are only accompanied by transcripts (ordered list of actions). Most of existing...
详细信息
暂无评论