28 accepted papers
Recently, the integration of large language models (LLMs) with generative recommendation (GR) has demonstrated promising potential. However, most existing GR methods adopt residual quantization to implicitly model hierarchical relationships across codebook layers in Euclidean space, which distorts t…
Domain generalization methods can effectively enhance network performance on test samples with unknown distributions by isolating gradients between unstable and stable parameters. However, existing methods employ relatively coarse-grained partitioning of stable versus unstable parameters, leading to…
Prototype-based methods enhance interpretability in image recognition by establishing intermediate part-level prototypes to build interpretable classifiers, enabling transparent decision-making through localized evidence and reference to prototypical examples. However, most existing methods typicall
Test-time domain generalization (TTDG) methods enhance the performance of neural networks on target domains by transferring the feature distribution of target samples to approximate that of the source domain, while avoiding the computational cost associated with fine-tuning on the target domain. How…
In the deep-learning-based computer vision community, Neural Architecture Search (NAS) has become the de-facto tool for acquiring task-optimal network structures. Nevertheless, NAS methods are trapped in a fundamental accuracy-efficiency dilemma: training-based approaches deliver reliable performanc
Data augmentation (DA) stands out as a powerful technique to enhance the generalization capabilities of deep neural networks across diverse tasks. However, in low-level vision tasks, DA remains rudimentary (i.e., vanilla DA), facing a critical bottleneck due to information loss. In this paper, we in…
Variable Subset Forecasting (VSF) presents a critical challenge where variable availability fluctuates between the training and inference stages. This paper introduces a novel solution to the underexplored VSF problem, seamlessly integrating with existing variate-wise attention architectures. Our ap…
Prototypical Part Networks facilitate interpretable decision-making processes, with the classification score computed by comparing test image patches to learned prototypes. Existing work typically learns prototypes from Convolutional Neural Networks (CNNs). However, learning prototypical parts direc…
The Mamba-type neural networks have gained significant popularity recently. To effectively and efficiently establish model architectures of Mamba, it is natural to introduce Neural Architecture Search (NAS) methods into Mamba. However, existing NAS methods tailored for Mamba are training-based, lead…
Data Augmentation (DA) is a powerful technique for enhancing the generalization capabilities of deep neural networks. It has been successfully employed in super-resolution research, resulting in improved performance. However, current DA methods often struggle to adapt effectively to light field supe…
The domain adaptation method effectively mitigates the negative impact of domain gaps on the performance of super-resolution (SR) networks through the guidance of numerous target domain low-resolution (LR) images. However, in real-world scenarios, the availability of target domain LR images is often…
In recent years, long-range time series forecasting has been actively studied and has shown promising results. However, since these methods mainly focus on predicting time series with a fixed dimension, they are inapplicable to the large-scale and ever-changing datasets that are common in real-world…
Zero-shot neural architecture search (NAS) has shown great potential in designing image recognition networks for its high efficiency and low resource consumption. However, most of the existing zero-shot NAS methods fail to utilize prior information in datasets when calculating the score of candidate…
Meta-learning has achieved remarkable success as a powerful paradigm for transferring learned knowledge from previous tasks. However, the lack of a large number of diverse and qualified tasks is the bottleneck of current meta-learning, which can easily lead to overfitting and therefore seriously hur…
Multi-hop question answering (QA) is a challenging task that requires complex reasoning over multiple documents. Existing works commonly introduce techniques such as graph modeling and question decomposition to explore precise intermediate results of multi-hop reasoning, leading to complexity growth…
DNN-based super-resolution(SR) models inherit the black-box nature of DNN and present low transparency. However, few works focus on interpreting low-level SR models and the limited existing gradient-based interpretability methods often require large computational costs and produce spurious/noisy fea…
Recently, Transformers have shown impressive performance in image super-resolution (SR), due to exploiting strong representation ability of multi-head self-attention (MSA). However, existing methods typically calculate MSA in a single range and granularity, preventing the model from capturing suffic…
Domain generalization (DG) aims to train a model on multiple source domains which can be well generalized to the unseen target domains. Currently, most DG techniques in vision tasks mainly focus on one of the three research lines (i.e., data manipulation, representation learning, and learning strate…
Graph Neural Networks (GNNs) have been a prevailing technique for tackling various analysis tasks on graph data. A key premise for the remarkable performance of GNNs relies on complete and trustworthy initial graph descriptions (i.e., node features and graph structure), which is often not satisfied…
Traffic prediction is an important component of the intelligent transportation system. Existing deep learning methods encode temporal information and spatial information separately or iteratively. However, the spatial and temporal information is highly correlated in a traffic network, so existing me…
This paper proposes ResTv2, a simpler, faster, and stronger multi-scale vision Transformer for visual recognition. ResTv2 simplifies the EMSA structure in ResTv1 (i.e., eliminating the multi-head interaction part) and employs an upsample operation to reconstruct the lost medium- and high-frequency i…
Existing part-based models for person Re-IDentification (Re-ID) usually suffer from part-misalignment problem caused by uniform partition of feature maps. The performances of part-based model are highly dependent on the semantically-aligned parts of the query and gallery images. However, misalignmen…
Deep neural networks have witnessed great success in Single Image Super-Resolution (SISR). However, current improvements are mainly contributed by much deeper networks, which leads to huge computation cost and limited application for mobile devices. Moreover, most existing methods propagate the basi…
This paper presents an efficient multi-scale vision Transformer, called ResT, that capably served as a general-purpose backbone for image recognition. Unlike existing Transformer methods, which employ standard Transformer blocks to tackle raw images with a fixed resolution, our ResT have several adv…
Attention mechanisms, which enable a neural network to accurately focus on all the relevant elements of the input, have become an essential component to improve the performance of deep neural networks. There are mainly two attention mechanisms widely used in computer vision studies, spatial attentio…
In recent years, part-based models have been verified their effectiveness for person Re-identification (Re-ID). Since they learn an embedding only by partitioning single-scale features of the highest layer in the backbone network, their performances highly depend on the well-aligned parts of the ext…
Convolutional neural networks (CNNs) have shown their power on many computer vision tasks. However, there are still some limitations, including their sensitivity to weight initialization and dependency to large scale labeled data. In this paper, we try to address these two problems by proposing a si…
In this paper, we propose a very deep fully convolutional encoding-decoding framework for image restoration such as denoising and super-resolution. The network is composed of multiple layers of convolution and deconvolution operators, learning end-to-end mappings from corrupted images to the origina…