Towards a Unified Analysis of Kernel-based Methods Under Covariate Shift

Covariate shift occurs prevalently in practice, where the input distributions of the source and target data are substantially different. Despite its practical importance in various learning problems, most of the existing methods only focus on some specific learning tasks and are not well validated theoretically and numerically. To tackle this problem, we propose a unified analysis of general nonparametric methods in a reproducing kernel Hilbert space (RKHS) under covariate shift. Our theoretical results are established for a general loss belonging to a rich loss function family, which includes many commonly used methods as special cases, such as mean regression, quantile regression, likelihood-based classification, and margin-based classification. Two types of covariate shift problems are the focus of this paper and the sharp convergence rates are established for a general loss function to provide a unified theoretical analysis, which concurs with the optimal results in literature where the squared loss is used. Extensive numerical studies on synthetic and real examples confirm our theoretical findings and further illustrate the effectiveness of our proposed method.

Optimize Weight Rounding via Signed Gradient Descent for the Quantization of LLMs

Large Language Models (LLMs) have proven their exceptional capabilities in performing language-related tasks. However, their deployment poses significant challenges due to their considerable memory and storage requirements. In response to this issue, weight-only quantization, particularly 3 and 4-bit weight-only quantization, has emerged as one of the most viable solutions. As the number of bits decreases, the quantization grid broadens, thus emphasizing the importance of up and down rounding. While previous studies have demonstrated that fine-tuning up and down rounding with the addition of perturbations can enhance accuracy in some scenarios, our study is driven by the precise and limited boundary of these perturbations, where only the threshold for altering the rounding value is of significance. Consequently, we propose a concise and highly effective approach for optimizing the weight rounding task. Our method, named SignRound, involves lightweight block-wise tuning using signed gradient descent, enabling us to achieve outstanding results within 400 steps. SignRound competes impressively against recent methods without introducing additional inference overhead. The source code will be publicly available at \url{https://github.com/intel/neural-compressor} soon.

Efficient Post-training Quantization with FP8 Formats

Recent advances in deep learning methods such as LLMs and Diffusion models have created a need for improved quantization methods that can meet the computational demands of these modern architectures while maintaining accuracy. Towards this goal, we study the advantages of FP8 data formats for post-training quantization across 75 unique network architectures covering a wide range of tasks, including machine translation, language modeling, text generation, image classification, generation, and segmentation. We examine three different FP8 representations (E5M2, E4M3, and E3M4) to study the effects of varying degrees of trade-off between dynamic range and precision on model accuracy. Based on our extensive study, we developed a quantization workflow that generalizes across different network architectures. Our empirical results show that FP8 formats outperform INT8 in multiple aspects, including workload coverage (92.64% vs. 65.87%), model accuracy and suitability for a broader range of operations. Furthermore, our findings suggest that E4M3 is better suited for NLP models, whereas E3M4 performs marginally better than E4M3 on computer vision tasks. The code is publicly available on Intel Neural Compressor: https://github.com/intel/neural-compressor.

FusionAI: Decentralized Training and Deploying LLMs with Massive Consumer-Level GPUs

The rapid growth of memory and computation requirements of large language models (LLMs) has outpaced the development of hardware, hindering people who lack large-scale high-end GPUs from training or deploying LLMs. However, consumer-level GPUs, which constitute a larger market share, are typically overlooked in LLM due to their weaker computing performance, smaller storage capacity, and lower communication bandwidth. Additionally, users may have privacy concerns when interacting with remote LLMs. In this paper, we envision a decentralized system unlocking the potential vast untapped consumer-level GPUs in pre-training, inference and fine-tuning of LLMs with privacy protection. However, this system faces critical challenges, including limited CPU and GPU memory, low network bandwidth, the variability of peer and device heterogeneity. To address these challenges, our system design incorporates: 1) a broker with backup pool to implement dynamic join and quit of computing providers; 2) task scheduling with hardware performance to improve system efficiency; 3) abstracting ML procedures into directed acyclic graphs (DAGs) to achieve model and task universality; 4) abstracting intermediate represention and execution planes to ensure compatibility of various devices and deep learning (DL) frameworks. Our performance analysis demonstrates that 50 RTX 3080 GPUs can achieve throughputs comparable to those of 4 H100 GPUs, which are significantly more expensive.

NeuralMatrix: Moving Entire Neural Networks to General Matrix Multiplication for Efficient Inference

In this study, we introduce NeuralMatrix, a novel framework that enables the computation of versatile deep neural networks (DNNs) on a single general matrix multiplication (GEMM) accelerator. The proposed approach overcomes the specificity limitations of ASIC-based accelerators while achieving application-specific acceleration levels compared to general-purpose processors such as CPUs and GPUs. We address the challenges of mapping both linear and nonlinear operations in DNN computation to general matrix multiplications and the impact of using a GEMM accelerator on DNN inference accuracy. Extensive experiments are conducted on various DNN models from three popular categories (i.e., CNN, Transformers, and GNN) as illustrative backbone models. Our results demonstrate that DNNs suffer only up to a 2.02% accuracy loss after being converted to general matrix multiplication, while achieving 113x to 19.44x improvements in throughput per power compared to CPUs and GPUs.

BS-GAT Behavior Similarity Based Graph Attention Network for Network Intrusion Detection

With the development of the Internet of Things (IoT), network intrusion detection is becoming more complex and extensive. It is essential to investigate an intelligent, automated, and robust network intrusion detection method. Graph neural networks based network intrusion detection methods have been proposed. However, it still needs further studies because the graph construction method of the existing methods does not fully adapt to the characteristics of the practical network intrusion datasets. To address the above issue, this paper proposes a graph neural network algorithm based on behavior similarity (BS-GAT) using graph attention network. First, a novel graph construction method is developed using the behavior similarity by analyzing the characteristics of the practical datasets. The data flows are treated as nodes in the graph, and the behavior rules of nodes are used as edges in the graph, constructing a graph with a relatively uniform number of neighbors for each node. Then, the edge behavior relationship weights are incorporated into the graph attention network to utilize the relationship between data flows and the structure information of the graph, which is used to improve the performance of the network intrusion detection. Finally, experiments are conducted based on the latest datasets to evaluate the performance of the proposed behavior similarity based graph attention network for the network intrusion detection. The results show that the proposed method is effective and has superior performance comparing to existing solutions.

Lformer: Text-to-Image Generation with L-shape Block Parallel Decoding

Generative transformers have shown their superiority in synthesizing high-fidelity and high-resolution images, such as good diversity and training stability. However, they suffer from the problem of slow generation since they need to generate a long token sequence autoregressively. To better accelerate the generative transformers while keeping good generation quality, we propose Lformer, a semi-autoregressive text-to-image generation model. Lformer firstly encodes an image into $h{\times}h$ discrete tokens, then divides these tokens into $h$ mirrored L-shape blocks from the top left to the bottom right and decodes the tokens in a block parallelly in each step. Lformer predicts the area adjacent to the previous context like autoregressive models thus it is more stable while accelerating. By leveraging the 2D structure of image tokens, Lformer achieves faster speed than the existing transformer-based methods while keeping good generation quality. Moreover, the pretrained Lformer can edit images without the requirement for finetuning. We can roll back to the early steps for regeneration or edit the image with a bounding box and a text prompt.

NAS-LID: Efficient Neural Architecture Search with Local Intrinsic Dimension

One-shot neural architecture search (NAS) substantially improves the search efficiency by training one supernet to estimate the performance of every possible child architecture (i.e., subnet). However, the inconsistency of characteristics among subnets incurs serious interference in the optimization, resulting in poor performance ranking correlation of subnets. Subsequent explorations decompose supernet weights via a particular criterion, e.g., gradient matching, to reduce the interference; yet they suffer from huge computational cost and low space separability. In this work, we propose a lightweight and effective local intrinsic dimension (LID)-based method NAS-LID. NAS-LID evaluates the geometrical properties of architectures by calculating the low-cost LID features layer-by-layer, and the similarity characterized by LID enjoys better separability compared with gradients, which thus effectively reduces the interference among subnets. Extensive experiments on NASBench-201 indicate that NAS-LID achieves superior performance with better efficiency. Specifically, compared to the gradient-driven method, NAS-LID can save up to 86% of GPU memory overhead when searching on NASBench-201. We also demonstrate the effectiveness of NAS-LID on ProxylessNAS and OFA spaces. Source code: https://github.com/marsggbo/NAS-LID.

Learning One-Class Hyperspectral Classifier from Positive and Unlabeled Data for Low Proportion Target

Hyperspectral imagery (HSI) one-class classification is aimed at identifying a single target class from the HSI by using only positive labels, which can significantly reduce the requirements for annotation. However, HSI one-class classification is far more challenging than HSI multi-class classification, due the lack of negative labels and the low target proportion, which are issues that have rarely been considered in the previous HSI classification studies. In this paper, a weakly supervised HSI one-class classifier, namely HOneCls is proposed to solve the problem of under-fitting of the positive class occurs in the HSI data with low target proportion, where a risk estimator -- the One-Class Risk Estimator -- is particularly introduced to make the full convolutional neural network (FCN) with the ability of one class classification. The experimental results obtained on challenging hyperspectral classification datasets, which includes 20 kinds of ground objects with very similar spectra, demonstrate the efficiency and feasibility of the proposed One-Class Risk Estimator. Compared with the state-of-the-art one-class classifiers, the F1-score is improved significantly in the HSI data with low target proportion.