Searcharxiv⌕ Search

arXiv subjects

Yi Ma

Publications and source records attributed to Yi Ma.

At least 163 records · Page 9Linked to original sources
arXiv2023

Scaff-PD: Communication Efficient Fair and Robust Federated Learning

We present Scaff-PD, a fast and communication-efficient algorithm for distributionally robust federated learning. Our approach improves fairness by optimizing a family of distributionally robust objectives tailored to heterogeneous clients. We leverage the special structure of these objectives, and design an accelerated primal dual (APD) algorithm which uses bias corrected local steps (as in Scaffold) to achieve significant gains in communication efficiency and convergence speed. We evaluate Scaff-PD on several benchmark datasets and demonstrate its effectiveness in improving fairness and robustness while maintaining competitive accuracy. Our results suggest that Scaff-PD is a promising approach for federated learning in resource-constrained and heterogeneous settings.

cs.LG↗
arXiv2023

Pixel-wise Graph Attention Networks for Person Re-identification

Graph convolutional networks (GCN) is widely used to handle irregular data since it updates node features by using the structure information of graph. With the help of iterated GCN, high-order information can be obtained to further enhance the representation of nodes. However, how to apply GCN to structured data (such as pictures) has not been deeply studied. In this paper, we explore the application of graph attention networks (GAT) in image feature extraction. First of all, we propose a novel graph generation algorithm to convert images into graphs through matrix transformation. It is one magnitude faster than the algorithm based on K Nearest Neighbors (KNN). Then, GAT is used on the generated graph to update the node features. Thus, a more robust representation is obtained. These two steps are combined into a module called pixel-wise graph attention module (PGA). Since the graph obtained by our graph generation algorithm can still be transformed into a picture after processing, PGA can be well combined with CNN. Based on these two modules, we consulted the ResNet and design a pixel-wise graph attention network (PGANet). The PGANet is applied to the task of person re-identification in the datasets Market1501, DukeMTMC-reID and Occluded-DukeMTMC (outperforms state-of-the-art by 0.8\%, 1.1\% and 11\% respectively, in mAP scores). Experiment results show that it achieves the state-of-the-art performance. \href{https://github.com/wenyu1009/PGANet}{The code is available here}.

cs.CV↗
arXiv2023

Achieving Maximum-likelihood Detection Performance with Square-order Complexity in Large Quasi-Symmetric MIMO Systems

We focus on the signal detection for large quasi-symmetric (LQS) multiple-input multiple-output (MIMO) systems, where the numbers of both service (M) and user (N) antennas are large and N/M tends to 1. It is challenging to achieve maximum-likelihood detection (MLD) performance with square-order complexity due to the ill-conditioned channel matrix. In the emerging MIMO paradigm termed with an extremely large aperture array, the channel matrix can be more ill-conditioned due to spatial non-stationarity. In this paper, projected-Jacobi (PJ) is proposed for signal detection in (non-) stationary LQS-MIMO systems. It is theoretically and empirically demonstrated that PJ can achieve MLD performance, even when N/M = 1. Moreover, PJ has square-order complexity of N and supports parallel computation. The main idea of PJ is to add a projection step and to set a (quasi-) orthogonal initialization for the classical Jacobi iteration. Moreover, the symbol error rate (SER) of PJ is mathematically derived and it is tight to the simulation results.

cs.IT↗
arXiv2023

ViP: A Differentially Private Foundation Model for Computer Vision

Artificial intelligence (AI) has seen a tremendous surge in capabilities thanks to the use of foundation models trained on internet-scale data. On the flip side, the uncurated nature of internet-scale data also poses significant privacy and legal risks, as they often contain personal information or copyrighted material that should not be trained on without permission. In this work, we propose as a mitigation measure a recipe to train foundation vision models with differential privacy (DP) guarantee. We identify masked autoencoders as a suitable learning algorithm that aligns well with DP-SGD, and train ViP -- a Vision transformer with differential Privacy -- under a strict privacy budget of $ε=8$ on the LAION400M dataset. We evaluate the quality of representation learned by ViP using standard downstream vision tasks; in particular, ViP achieves a (non-private) linear probing accuracy of $55.7\%$ on ImageNet, comparable to that of end-to-end trained AlexNet (trained and evaluated on ImageNet). Our result suggests that scaling to internet-scale data can be practical for private learning. Code is available at \url{https://github.com/facebookresearch/ViP-MAE}.

cs.CV↗
arXiv2023

Understanding the Complexity Gains of Single-Task RL with a Curriculum

Reinforcement learning (RL) problems can be challenging without well-shaped rewards. Prior work on provably efficient RL methods generally proposes to address this issue with dedicated exploration strategies. However, another way to tackle this challenge is to reformulate it as a multi-task RL problem, where the task space contains not only the challenging task of interest but also easier tasks that implicitly function as a curriculum. Such a reformulation opens up the possibility of running existing multi-task RL methods as a more efficient alternative to solving a single challenging task from scratch. In this work, we provide a theoretical framework that reformulates a single-task RL problem as a multi-task RL problem defined by a curriculum. Under mild regularity conditions on the curriculum, we show that sequentially solving each task in the multi-task RL problem is more computationally efficient than solving the original single-task problem, without any explicit exploration bonuses or other exploration strategies. We also show that our theoretical insights can be translated into an effective practical learning algorithm that can accelerate curriculum learning on simulated robotic tasks.

cs.LG↗
arXiv2023

HIPODE: Enhancing Offline Reinforcement Learning with High-Quality Synthetic Data from a Policy-Decoupled Approach

Offline reinforcement learning (ORL) has gained attention as a means of training reinforcement learning models using pre-collected static data. To address the issue of limited data and improve downstream ORL performance, recent work has attempted to expand the dataset's coverage through data augmentation. However, most of these methods are tied to a specific policy (policy-dependent), where the generated data can only guarantee to support the current downstream ORL policy, limiting its usage scope on other downstream policies. Moreover, the quality of synthetic data is often not well-controlled, which limits the potential for further improving the downstream policy. To tackle these issues, we propose \textbf{HI}gh-quality \textbf{PO}licy-\textbf{DE}coupled~(HIPODE), a novel data augmentation method for ORL. On the one hand, HIPODE generates high-quality synthetic data by selecting states near the dataset distribution with potentially high value among candidate states using the negative sampling technique. On the other hand, HIPODE is policy-decoupled, thus can be used as a common plug-in method for any downstream ORL process. We conduct experiments on the widely studied TD3BC and CQL algorithms, and the results show that HIPODE outperforms the state-of-the-art policy-decoupled data augmentation method and most prevalent model-based ORL methods on D4RL benchmarks.

cs.LG↗
arXiv2023

Incremental Learning of Structured Memory via Closed-Loop Transcription

This work proposes a minimal computational model for learning structured memories of multiple object classes in an incremental setting. Our approach is based on establishing a closed-loop transcription between the classes and a corresponding set of subspaces, known as a linear discriminative representation, in a low-dimensional feature space. Our method is simpler than existing approaches for incremental learning, and more efficient in terms of model size, storage, and computation: it requires only a single, fixed-capacity autoencoding network with a feature space that is used for both discriminative and generative purposes. Network parameters are optimized simultaneously without architectural manipulations, by solving a constrained minimax game between the encoding and decoding maps over a single rate reduction-based objective. Experimental results show that our method can effectively alleviate catastrophic forgetting, achieving significantly better performance than prior work of generative replay on MNIST, CIFAR-10, and ImageNet-50, despite requiring fewer resources. Source code can be found at https://github.com/tsb0601/i-CTRL

cs.CV↗
arXiv2023

White-Box Transformers via Sparse Rate Reduction

In this paper, we contend that the objective of representation learning is to compress and transform the distribution of the data, say sets of tokens, towards a mixture of low-dimensional Gaussian distributions supported on incoherent subspaces. The quality of the final representation can be measured by a unified objective function called sparse rate reduction. From this perspective, popular deep networks such as transformers can be naturally viewed as realizing iterative schemes to optimize this objective incrementally. Particularly, we show that the standard transformer block can be derived from alternating optimization on complementary parts of this objective: the multi-head self-attention operator can be viewed as a gradient descent step to compress the token sets by minimizing their lossy coding rate, and the subsequent multi-layer perceptron can be viewed as attempting to sparsify the representation of the tokens. This leads to a family of white-box transformer-like deep network architectures which are mathematically fully interpretable. Despite their simplicity, experiments show that these networks indeed learn to optimize the designed objective: they compress and sparsify representations of large-scale real-world vision datasets such as ImageNet, and achieve performance very close to thoroughly engineered transformers such as ViT. Code is at \url{https://github.com/Ma-Lab-Berkeley/CRATE}.

cs.LG↗
arXiv2023

Space MIMO: Direct Unmodified Handheld to Multi-Satellite Communication

This paper examines the uplink transmission of a single-antenna handsheld user to a cluster of satellites, with a focus on utilizing the inter-satellite links to enable cooperative signal detection. Two cases are studied: one with full CSI and the other with partial CSI between satellites. The two cases are compared in terms of capacity, overhead, and bit error rate. Additionally, the impact of channel estimation error is analyzed in both designs, and robust detection techniques are proposed to handle channel uncertainty up to a certain level. The performance of each case is demonstrated, and a comparison is made with conventional satellite communication schemes where only one satellite can connect to a user. The results of our study reveal that the proposed constellation with a total of 3168 satellites in orbit can enable a capacity of 800 Mbits/sec through cooperation of $12$ satellites with and occupied bandwidth of 500 MHz. In contrast, conventional satellite communication approaches with the same system parameters yield a significantly lower capacity of less than 150 Mbits/sec for the nearest satellite.

cs.IT↗
arXiv2023

Minimalistic Unsupervised Learning with the Sparse Manifold Transform

We describe a minimalistic and interpretable method for unsupervised learning, without resorting to data augmentation, hyperparameter tuning, or other engineering designs, that achieves performance close to the SOTA SSL methods. Our approach leverages the sparse manifold transform, which unifies sparse coding, manifold learning, and slow feature analysis. With a one-layer deterministic sparse manifold transform, one can achieve 99.3% KNN top-1 accuracy on MNIST, 81.1% KNN top-1 accuracy on CIFAR-10 and 53.2% on CIFAR-100. With a simple gray-scale augmentation, the model gets 83.2% KNN top-1 accuracy on CIFAR-10 and 57% on CIFAR-100. These results significantly close the gap between simplistic "white-box" methods and the SOTA methods. Additionally, we provide visualization to explain how an unsupervised representation transform is formed. The proposed method is closely connected to latent-embedding self-supervised methods and can be treated as the simplest form of VICReg. Though there remains a small performance gap between our simple constructive model and SOTA methods, the evidence points to this as a promising direction for achieving a principled and white-box approach to unsupervised learning.

cs.LG↗
arXiv2023

EMP-SSL: Towards Self-Supervised Learning in One Training Epoch

Recently, self-supervised learning (SSL) has achieved tremendous success in learning image representation. Despite the empirical success, most self-supervised learning methods are rather "inefficient" learners, typically taking hundreds of training epochs to fully converge. In this work, we show that the key towards efficient self-supervised learning is to increase the number of crops from each image instance. Leveraging one of the state-of-the-art SSL method, we introduce a simplistic form of self-supervised learning method called Extreme-Multi-Patch Self-Supervised-Learning (EMP-SSL) that does not rely on many heuristic techniques for SSL such as weight sharing between the branches, feature-wise normalization, output quantization, and stop gradient, etc, and reduces the training epochs by two orders of magnitude. We show that the proposed method is able to converge to 85.1% on CIFAR-10, 58.5% on CIFAR-100, 38.1% on Tiny ImageNet and 58.5% on ImageNet-100 in just one epoch. Furthermore, the proposed method achieves 91.5% on CIFAR-10, 70.1% on CIFAR-100, 51.5% on Tiny ImageNet and 78.9% on ImageNet-100 with linear probing in less than ten training epochs. In addition, we show that EMP-SSL shows significantly better transferability to out-of-domain datasets compared to baseline SSL methods. We will release the code in https://github.com/tsb0601/EMP-SSL.

cs.CV↗
arXiv2023

Temperature and Pressure-Induced Atomic Structure Evolution During Solidification of Zr50Nb50 Metallic Melt via Molecular Dynamics Simulation

In this report, the evolution of the local atomic structure of the Zr50Nb50 melt was investigated by applying temperature (2600 to 300 K) and pressure (0 to 50 Gpa) using classical molecular dynamics simulations. To gain clear insight into the structural evolution during quenching, we used various methods of structural analysis such as the radial distribution function g(r), coordination number, bond angle distribution, and Voronoi tessellation. We found that the icosahedral motifs (which are the signature of the short-range ordering) and distorted BCC-like clusters dominate in the liquid and glass region under 0 and 5 Gpa external pressure. A first-order phase transition to a crystal-like structure was observed at 10, 15, and 20 Gpa external pressure at 1400, 1500, and 1600 K, respectively. Before the first-order phase transition, the system was dominated by icosahedral and distorted BCC-like clusters. When the temperature is lowered further below the glass transition at 10,15, and 20 Gpa external pressure, all structural analyses show that the solidified system consists mainly of body-centered cubic-like clusters in the case of our specific cooling rate of 1012 K/s.

cond-mat.mtrl-sci↗
arXiv2023

Closed-Loop Transcription via Convolutional Sparse Coding

Autoencoding has achieved great empirical success as a framework for learning generative models for natural images. Autoencoders often use generic deep networks as the encoder or decoder, which are difficult to interpret, and the learned representations lack clear structure. In this work, we make the explicit assumption that the image distribution is generated from a multi-stage sparse deconvolution. The corresponding inverse map, which we use as an encoder, is a multi-stage convolution sparse coding (CSC), with each stage obtained from unrolling an optimization algorithm for solving the corresponding (convexified) sparse coding program. To avoid computational difficulties in minimizing distributional distance between the real and generated images, we utilize the recent closed-loop transcription (CTRL) framework that optimizes the rate reduction of the learned sparse representations. Conceptually, our method has high-level connections to score-matching methods such as diffusion models. Empirically, our framework demonstrates competitive performance on large-scale datasets, such as ImageNet-1K, compared to existing autoencoding and generative methods under fair conditions. Even with simpler networks and fewer computational resources, our method demonstrates high visual quality in regenerated images. More surprisingly, the learned autoencoder performs well on unseen datasets. Our method enjoys several side benefits, including more structured and interpretable representations, more stable convergence, and scalability to large datasets. Our method is arguably the first to demonstrate that a concatenation of multiple convolution sparse coding/decoding layers leads to an interpretable and effective autoencoder for modeling the distribution of large-scale natural image datasets.

cs.CV↗
arXiv2023

Sherman-Morrison Regularization for ELAA Iterative Linear Precoding

The design of iterative linear precoding is recently challenged by extremely large aperture array (ELAA) systems, where conventional preconditioning techniques could hardly improve the channel condition. In this paper, it is proposed to regularize the extreme singular values to improve the channel condition by deducting a rank-one matrix from the Wishart matrix of the channel. Our analysis proves the feasibility to reduce the largest singular value or to increase multiple small singular values with a rank-one matrix when the singular value decomposition of the channel is available. Knowing the feasibility, we propose a low-complexity approach where an approximation of the regularization matrix can be obtained based on the statistical property of the channel. It is demonstrated, through simulation results, that the proposed low-complexity approach significantly outperforms current preconditioning techniques in terms of reduced iteration number for more than $10\%$ in both ELAA systems as well as symmetric multi-antenna (i.e., MIMO) systems when the channel is i.i.d. Rayleigh fading.

eess.SP↗
arXiv2023

Robot Subset Selection for Swarm Lifetime Maximization in Computation Offloading with Correlated Data Sources

Consider robot swarm wireless networks where mobile robots offload their computing tasks to a computing server located at the mobile edge. Our aim is to maximize the swarm lifetime through efficient exploitation of the correlation between distributed data sources. The optimization problem is handled by selecting appropriate robot subsets to send their sensed data to the server. In this work, the data correlation between distributed robot subsets is modelled as an undirected graph. A least-degree iterative partitioning (LDIP) algorithm is proposed to partition the graph into a set of subgraphs. Each subgraph has at least one vertex (i.e., subset), termed representative vertex (R-Vertex), which shares edges with and only with all other vertices within the subgraph; only R-Vertices are selected for data transmissions. When the number of subgraphs is maximized, the proposed subset selection approach is shown to be optimum in the AWGN channel. For independent fading channels, the max-min principle can be incorporated into the proposed approach to achieve the best performance.

eess.SY↗
arXiv2022

Fully Convolutional Line Parsing

We present a one-stage Fully Convolutional Line Parsing network (F-Clip) that detects line segments from images. The proposed network is very simple and flexible with variations that gracefully trade off between speed and accuracy for different applications. F-Clip detects line segments in an end-to-end fashion by predicting each line's center position, length, and angle. We further customize the design of convolution kernels of our fully convolutional network to effectively exploit the statistical priors of the distribution of line angles in real image datasets. We conduct extensive experiments and show that our method achieves a significantly better trade-off between efficiency and accuracy, resulting in a real-time line detector at up to 73 FPS on a single GPU. Such inference speed makes our method readily applicable to real-time tasks without compromising any accuracy of previous methods. Moreover, when equipped with a performance-improving backbone network, F-Clip is able to significantly outperform all state-of-the-art line detectors on accuracy at a similar or even higher frame rate. In other word, under same inference speed, F-Clip always achieving best accuracy compare with other methods. Source code https://github.com/Delay-Xili/F-Clip.

cs.CV↗
arXiv2022

Disparity-based Stereo Image Compression with Aligned Cross-View Priors

With the wide application of stereo images in various fields, the research on stereo image compression (SIC) attracts extensive attention from academia and industry. The core of SIC is to fully explore the mutual information between the left and right images and reduce redundancy between views as much as possible. In this paper, we propose DispSIC, an end-to-end trainable deep neural network, in which we jointly train a stereo matching model to assist in the image compression task. Based on the stereo matching results (i.e. disparity), the right image can be easily warped to the left view, and only the residuals between the left and right views are encoded for the left image. A three-branch auto-encoder architecture is adopted in DispSIC, which encodes the right image, the disparity map and the residuals respectively. During training, the whole network can learn how to adaptively allocate bitrates to these three parts, achieving better rate-distortion performance at the cost of a lower disparity map bitrates. Moreover, we propose a conditional entropy model with aligned cross-view priors for SIC, which takes the warped latents of the right image as priors to improve the accuracy of the probability estimation for the left image. Experimental results demonstrate that our proposed method achieves superior performance compared to other existing SIC methods on the KITTI and InStereo2K datasets both quantitatively and qualitatively.

cs.MM↗
arXiv2022

State-Aware Proximal Pessimistic Algorithms for Offline Reinforcement Learning

Pessimism is of great importance in offline reinforcement learning (RL). One broad category of offline RL algorithms fulfills pessimism by explicit or implicit behavior regularization. However, most of them only consider policy divergence as behavior regularization, ignoring the effect of how the offline state distribution differs with that of the learning policy, which may lead to under-pessimism for some states and over-pessimism for others. Taking account of this problem, we propose a principled algorithmic framework for offline RL, called \emph{State-Aware Proximal Pessimism} (SA-PP). The key idea of SA-PP is leveraging discounted stationary state distribution ratios between the learning policy and the offline dataset to modulate the degree of behavior regularization in a state-wise manner, so that pessimism can be implemented in a more appropriate way. We first provide theoretical justifications on the superiority of SA-PP over previous algorithms, demonstrating that SA-PP produces a lower suboptimality upper bound in a broad range of settings. Furthermore, we propose a new algorithm named \emph{State-Aware Conservative Q-Learning} (SA-CQL), by building SA-PP upon representative CQL algorithm with the help of DualDICE for estimating discounted stationary state distribution ratios. Extensive experiments on standard offline RL benchmark show that SA-CQL outperforms the popular baselines on a large portion of benchmarks and attains the highest average return.

cs.LG↗