Li Li

Liang Gao, Huazhu Fu, Li Li et al.

Federated learning (FL) allows multiple clients to collectively train a high-performance global model without sharing their private data. However, the key challenge in federated learning is that the clients have significant statistical heterogeneity among their local data distributions, which would cause inconsistent optimized local models on the client-side. To address this fundamental dilemma, we propose a novel federated learning algorithm with local drift decoupling and correction (FedDC). Our FedDC only introduces lightweight modifications in the local training phase, in which each client utilizes an auxiliary local drift variable to track the gap between the local model parameter and the global model parameters. The key idea of FedDC is to utilize this learned local drift variable to bridge the gap, i.e., conducting consistency in parameter-level. The experiment results and analysis demonstrate that FedDC yields expediting convergence and better performance on various image classification tasks, robust in partial participation settings, non-iid data, and heterogeneous clients.

Yonghui Wang, Wengang Zhou, Hao Feng et al.

Single-image shadow removal is a significant task that is still unresolved. Most existing deep learning-based approaches attempt to remove the shadow directly, which can not deal with the shadow well. To handle this issue, we consider removing the shadow in a coarse-to-fine fashion and propose a simple but effective Progressive Recurrent Network (PRNet). The network aims to remove the shadow progressively, enabing us to flexibly adjust the number of iterations to strike a balance between performance and time. Our network comprises two parts: shadow feature extraction and progressive shadow removal. Specifically, the first part is a shallow ResNet which constructs the representations of the input shadow image on its original size, preventing the loss of high-frequency details caused by the downsampling operation. The second part has two critical components: the re-integration module and the update module. The proposed re-integration module can fully use the outputs of the previous iteration, providing input for the update module for further shadow removal. In this way, the proposed PRNet makes the whole process more concise and only uses 29% network parameters than the best published method. Extensive experiments on the three benchmarks, ISTD, ISTD+, and SRD, demonstrate that our method can effectively remove shadows and achieve superior performance.

Keyi Zhou, Li Li, Wengang Zhou et al.

In video lane detection, there are rich temporal contexts among successive frames, which is under-explored in existing lane detectors. In this work, we propose LaneTCA to bridge the individual video frames and explore how to effectively aggregate the temporal context. Technically, we develop an accumulative attention module and an adjacent attention module to abstract the long-term and short-term temporal context, respectively. The accumulative attention module continuously accumulates visual information during the journey of a vehicle, while the adjacent attention module propagates this lane information from the previous frame to the current frame. The two modules are meticulously designed based on the transformer architecture. Finally, these long-short context features are fused with the current frame features to predict the lane lines in the current frame. Extensive quantitative and qualitative experiments are conducted on two prevalent benchmark datasets. The results demonstrate the effectiveness of our method, achieving several new state-of-the-art records. The codes and models are available at https://github.com/Alex-1337/LaneTCA

Jiawei Zhang, Shen Li, Li Li

Connected and automated vehicles (CAVs) are viewed as a special kind of robots that have the potential to significantly improve the safety and efficiency of traffic. In contrast to many swarm robotics studies that are demonstrated in labs by employing a small number of robots, CAV studies aims to achieve cooperative driving of unceasing robot swarm flows. However, how to get the optimal passing order of such robot swarm flows even for a signal-free intersection is an NP-hard problem (specifically, enumerating based algorithm takes days to find the optimal solution to a 20-CAV scenario). Here, we introduce a novel cooperative driving algorithm (AlphaOrder) that combines offline deep learning and online tree searching to find a near-optimal passing order in real-time. AlphaOrder builds a pointer network model from solved scenarios and generates near-optimal passing orders instantaneously for new scenarios. Furthermore, our approach provides a general approach to managing preemptive resource sharing between swarm robotics (e.g., scheduling multiple automated guided vehicles (AGVs) and unmanned aerial vehicles (UAVs) at conflicting areas

Dachong Li, Li Li, Zhuangzhuang Chen et al.

Large kernels make standard convolutional neural networks (CNNs) great again over transformer architectures in various vision tasks. Nonetheless, recent studies meticulously designed around increasing kernel size have shown diminishing returns or stagnation in performance. Thus, the hidden factors of large kernel convolution that affect model performance remain unexplored. In this paper, we reveal that the key hidden factors of large kernels can be summarized as two separate components: extracting features at a certain granularity and fusing features by multiple pathways. To this end, we leverage the multi-path long-distance sparse dependency relationship to enhance feature utilization via the proposed Shiftwise (SW) convolution operator with a pure CNN architecture. In a wide range of vision tasks such as classification, segmentation, and detection, SW surpasses state-of-the-art transformers and CNN architectures, including SLaK and UniRepLKNet. More importantly, our experiments demonstrate that $3 \times 3$ convolutions can replace large convolutions in existing large kernel CNNs to achieve comparable effects, which may inspire follow-up works. Code and all the models at https://github.com/lidc54/shift-wiseConv.

Xiangquan Gui, Binxuan Zhang, Li Li et al.

Chinese landscape painting has a unique and artistic style, and its drawing technique is highly abstract in both the use of color and the realistic representation of objects. Previous methods focus on transferring from modern photos to ancient ink paintings. However, little attention has been paid to translating landscape paintings into modern photos. To solve such problems, in this paper, we (1) propose DLP-GAN (Draw Modern Chinese Landscape Photos with Generative Adversarial Network), an unsupervised cross-domain image translation framework with a novel asymmetric cycle mapping, and (2) introduce a generator based on a dense-fusion module to match different translation directions. Moreover, a dual-consistency loss is proposed to balance the realism and abstraction of model painting. In this way, our model can draw landscape photos and sketches in the modern sense. Finally, based on our collection of modern landscape and sketch datasets, we compare the images generated by our model with other benchmarks. Extensive experiments including user studies show that our model outperforms state-of-the-art methods.

Jiazhang Zheng, Lei Li, Qiuping Liao et al. · pku

Nighttime photography encounters escalating challenges in extremely low-light conditions, primarily attributable to the ultra-low signal-to-noise ratio. For real-world deployment, a practical solution must not only produce visually appealing results but also require minimal computation. However, most existing methods are either focused on improving restoration performance or employ lightweight models at the cost of quality. This paper proposes a lightweight network that outperforms existing state-of-the-art (SOTA) methods in low-light enhancement tasks while minimizing computation. The proposed network incorporates Siamese Self-Attention Block (SSAB) and Skip-Channel Attention (SCA) modules, which enhance the model's capacity to aggregate global information and are well-suited for high-resolution images. Additionally, based on our analysis of the low-light image restoration process, we propose a Two-Stage Framework that achieves superior results. Our model can restore a UHD 4K resolution image with minimal computation while keeping SOTA restoration quality.

Binchuan Qi, Wei Gong, Li Li

Despite its empirical success, deep learning still lacks a comprehensive theoretical understanding of model fitting and generalization. This paper proposes the probability distribution (PD) learning framework to analyze the optimization and generalization mechanisms of deep learning. Within this framework, the conditional distribution of labels given features is the primary learning target, with the loss function, prior knowledge, and model properties explicitly characterized. Under these formulations, we establish theoretical guarantees on optimizability, even in non-convex settings, and derive generalization error bounds that provide meaningful explanations for practical performance. Specifically, we first prove theoretically that the Fenchel-Young loss is the natural and necessary choice for solving PD learning problems, thereby justifying the generality of conclusions based on this loss. Second, to capture the characteristics of deep neural networks (DNNs), we introduce the notions of $\mathcal{H}(ψ)$-convexity and $\mathcal{H}(Ψ)$-smoothness, which generalize the classical concepts of strong convexity and Lipschitz smoothness. Based on them, we provide a theoretical explanation for the effectiveness of SGD in training DNNs. Finally, we derive model-independent bounds on the expected risk and generalization error for trained models, revealing the influence of the training set size, regularization term, the mutual information between labels and features, and the information loss caused by model irreversibility on risk and generalization. Based on our theoretical analysis and experimental validation, we believe that the PD learning framework facilitates a deeper and more unified theoretical understanding of deep learning.

Zhensu Sun, Xiaoning Du, Fu Song et al.

Leveraging recent advancements in large language models, modern neural code completion models have demonstrated the capability to generate highly accurate code suggestions. However, their massive size poses challenges in terms of computational costs and environmental impact, hindering their widespread adoption in practical scenarios. Dynamic inference emerges as a promising solution, as it allocates minimal computation during inference while maintaining the model's performance. In this research, we explore dynamic inference within the context of code completion. Initially, we conducted an empirical investigation on GPT-2, focusing on the inference capabilities of intermediate layers for code completion. We found that 54.4% of tokens can be accurately generated using just the first layer, signifying significant computational savings potential. Moreover, despite using all layers, the model still fails to predict 14.5% of tokens correctly, and the subsequent completions continued from them are rarely considered helpful, with only a 4.2% Acceptance Rate. These findings motivate our exploration of dynamic inference in code completion and inspire us to enhance it with a decision-making mechanism that stops the generation of incorrect code. We thus propose a novel dynamic inference method specifically tailored for code completion models. This method aims not only to produce correct predictions with largely reduced computation but also to prevent incorrect predictions proactively. Our extensive evaluation shows that it can averagely skip 1.7 layers out of 16 layers in the models, leading to an 11.2% speedup with only a marginal 1.1% reduction in ROUGE-L.

Yangjie Zhou, Yaoxu Song, Jingwen Leng et al.

Graph neural networks (GNNs) are powerful tools for exploring and learning from graph structures and features. As such, achieving high-performance execution for GNNs becomes crucially important. Prior works have proposed to explore the sparsity (i.e., low density) in the input graph to accelerate GNNs, which uses the full-graph-level or block-level sparsity format. We show that they fail to balance the sparsity benefit and kernel execution efficiency. In this paper, we propose a novel system, referred to as AdaptGear, that addresses the challenge of optimizing GNNs performance by leveraging kernels tailored to the density characteristics at the subgraph level. Meanwhile, we also propose a method that dynamically chooses the optimal set of kernels for a given input graph. Our evaluation shows that AdaptGear can achieve a significant performance improvement, up to $6.49 \times$ ($1.87 \times$ on average), over the state-of-the-art works on two mainstream NVIDIA GPUs across various datasets.

Xiuxian Li, Kuo-Yi Lin, Li Li et al.

Communication has been seen as a significant bottleneck in industrial applications over large-scale networks. To alleviate the communication burden, sign-based optimization algorithms have gained popularity recently in both industrial and academic communities, which is shown to be closely related to adaptive gradient methods, such as Adam. Along this line, this paper investigates faster convergence for a variant of sign-based gradient descent, called scaled signGD, in three cases: 1) the objective function is strongly convex; 2) the objective function is nonconvex but satisfies the Polyak-Lojasiewicz (PL) inequality; 3) the gradient is stochastic, called scaled signGD in this case. For the first two cases, it can be shown that the scaled signGD converges at a linear rate. For case 3), the algorithm is shown to converge linearly to a neighborhood of the optimal value when a constant learning rate is employed, and the algorithm converges at a rate of $O(1/k)$ when using a diminishing learning rate, where $k$ is the iteration number. The results are also extended to the distributed setting by majority vote in a parameter-server framework. Finally, numerical experiments on logistic regression are performed to corroborate the theoretical findings.

Xinran Li, Kuo-Yi Lin, Min Meng et al.

Due to the growing awareness of driving safety and the development of sophisticated technologies, advanced driving assistance system (ADAS) has been equipped in more and more vehicles with higher accuracy and lower price. The latest progress in this field has called for a review to sum up the conventional knowledge of ADAS, the state-of-the-art researches, and novel applications in real-world. With the help of this kind of review, newcomers in this field can get basic knowledge easier and other researchers may be inspired with potential future development possibility. This paper makes a general introduction about ADAS by analyzing its hardware support and computation algorithms. Different types of perception sensors are introduced from their interior feature classifications, installation positions, supporting ADAS functions, and pros and cons. The comparisons between different sensors are concluded and illustrated from their inherent characters and specific usages serving for each ADAS function. The current algorithms for ADAS functions are also collected and briefly presented in this paper from both traditional methods and novel ideas. Additionally, discussions about the definition of ADAS from different institutes are reviewed in this paper, and future approaches about ADAS in China are introduced in particular.

Xinyu Peng, Jiawei Zhang, Fei-Yue Wang et al.

Deep learning has become the most powerful machine learning tool in the last decade. However, how to efficiently train deep neural networks remains to be thoroughly solved. The widely used minibatch stochastic gradient descent (SGD) still needs to be accelerated. As a promising tool to better understand the learning dynamic of minibatch SGD, the information bottleneck (IB) theory claims that the optimization process consists of an initial fitting phase and the following compression phase. Based on this principle, we further study typicality sampling, an efficient data selection method, and propose a new explanation of how it helps accelerate the training process of the deep networks. We show that the fitting phase depicted in the IB theory will be boosted with a high signal-to-noise ratio of gradient approximation if the typicality sampling is appropriately adopted. Furthermore, this finding also implies that the prior information of the training set is critical to the optimization process and the better use of the most important data can help the information flow through the bottleneck faster. Both theoretical analysis and experimental results on synthetic and real-world datasets demonstrate our conclusions.

Fanghui Liu, Xiaolin Huang, Chen Gong et al.

In this paper, we propose a data-adaptive non-parametric kernel learning framework in margin based kernel methods. In model formulation, given an initial kernel matrix, a data-adaptive matrix with two constraints is imposed in an entry-wise scheme. Learning this data-adaptive matrix in a formulation-free strategy enlarges the margin between classes and thus improves the model flexibility. The introduced two constraints are imposed either exactly (on small data sets) or approximately (on large data sets) in our model, which provides a controllable trade-off between model flexibility and complexity with theoretical demonstration. In algorithm optimization, the objective function of our learning framework is proven to be gradient-Lipschitz continuous. Thereby, kernel and classifier/regressor learning can be efficiently optimized in a unified framework via Nesterov's acceleration. For the scalability issue, we study a decomposition-based approach to our model in the large sample case. The effectiveness of this approximation is illustrated by both empirical studies and theoretical guarantees. Experimental results on various classification and regression benchmark data sets demonstrate that our non-parametric kernel learning framework achieves good performance when compared with other representative kernel learning based algorithms.

Michael Wojnowicz, Dinh Nguyen, Li Li et al.

Stochastic principal component analysis (SPCA) has become a popular dimensionality reduction strategy for large, high-dimensional datasets. We derive a simplified algorithm, called Lazy SPCA, which has reduced computational complexity and is better suited for large-scale distributed computation. We prove that SPCA and Lazy SPCA find the same approximations to the principal subspace, and that the pairwise distances between samples in the lower-dimensional space is invariant to whether SPCA is executed lazily or not. Empirical studies find downstream predictive performance to be identical for both methods, and superior to random projections, across a range of predictive models (linear regression, logistic lasso, and random forests). In our largest experiment with 4.6 million samples, Lazy SPCA reduced 43.7 hours of computation to 9.9 hours. Overall, Lazy SPCA relies exclusively on matrix multiplications, besides an operation on a small square matrix whose size depends only on the target dimensionality.