Searcharxiv⌕ Search

arXiv subjects

Jie Zhang

Publications and source records attributed to Jie Zhang.

At least 703 records · Page 39Linked to original sources
arXiv2023

Towards Omni-generalizable Neural Methods for Vehicle Routing Problems

Learning heuristics for vehicle routing problems (VRPs) has gained much attention due to the less reliance on hand-crafted rules. However, existing methods are typically trained and tested on the same task with a fixed size and distribution (of nodes), and hence suffer from limited generalization performance. This paper studies a challenging yet realistic setting, which considers generalization across both size and distribution in VRPs. We propose a generic meta-learning framework, which enables effective training of an initialized model with the capability of fast adaptation to new tasks during inference. We further develop a simple yet efficient approximation method to reduce the training overhead. Extensive experiments on both synthetic and benchmark instances of the traveling salesman problem (TSP) and capacitated vehicle routing problem (CVRP) demonstrate the effectiveness of our method. The code is available at: https://github.com/RoyalSkye/Omni-VRP.

cs.LG↗
arXiv2023

A Survey on User-Space Storage and Its Implementations

The storage stack in the traditional operating system is primarily optimized towards improving the CPU utilization and hiding the long I/O latency imposed by the slow I/O devices such as hard disk drivers (HDDs). However, the emerging storage media experience significant technique shifts in the past decade, which exhibit high bandwidth and low latency. These high-performance storage devices, unfortunately, suffer from the huge overheads imposed by the system software including the long storage stack and the frequent context switch between the user and kernel modes. Many researchers have investigated huge efforts in addressing this challenge by constructing a direct software path between a user process and the underlying storage devices. We revisit such novel designs in the prior work and present a survey in this paper. Specifically, we classify the former research into three categories according to their commonalities. We then present the designs of each category based on the timeline and analyze their uniqueness and contributions. This paper also reviews the applications that exploit the characteristics of theses designs. Given that the user-space storage is a growing research field, we believe this paper can be an inspiration for future researchers, who are interested in the user-space storage system designs.

cs.OS↗
arXiv2023

Legion: Automatically Pushing the Envelope of Multi-GPU System for Billion-Scale GNN Training

Graph neural network(GNN) has been widely applied in real-world applications, such as product recommendation in e-commerce platforms and risk control in financial management systems. Several cache-based GNN systems have been built to accelerate GNN training in a single machine with multiple GPUs. However, these systems fail to train billion-scale graphs efficiently, which is a common challenge in the industry. In this work, we propose Legion, a system that automatically pushes the envelope of multi-GPU systems for accelerating billion-scale GNN training. First, we design a hierarchical graph partitioning mechanism that significantly improves the multi-GPU cache performance. Second, we build a unified multi-GPU cache that helps to minimize the PCIe traffic incurred by caching both graph topology and features with the highest hotness. Third, we develop an automatic caching management mechanism that adapts the multi-GPU cache plan according to the hardware specifications and various graphs to maximize the overall training throughput. Evaluations on various GNN models and multiple datasets show that Legion supports training billion-scale GNNs in a single machine and significantly outperforms the state-of-the-art cache-based systems on small graphs.

cs.DC↗
arXiv2023

Phonon dynamic behaviors induced by amorphous interlayer at heterointerfaces

Interface impedes heat flow in heterostructures and the interfacial thermal resistance (ITR) has become a critical issue for thermal dissipation in electronic devices. To explore the mechanism leading to the ITR, in this work, the dynamic behaviors of phonons passing through the GaN/AlN interface with an amorphous interlayer is investigated by using phonon wave packet simulation. It is found the amorphous interlayer significantly impedes phonon transport across the interface, and leads to remarkable phonon mode conversions, such as LA$\rightarrow$TA, TA$\rightarrow$LA, and LA$\rightarrow$TO conversion. However, due to mode conversion and inelastic scattering, we found a portion of high-frequency TA phonons, which are higher than the cut-off frequency and cannot transmit across the ideal sharp interface, can partially transmit across the amorphous interlayer, which introduces additional thermal transport channels through the interface and has positive effect on interfacial thermal conductance. According to phonon transmission coefficient, it is found the ITR increases with increasing of amorphous interlayer thickness L. The phonon transmission coefficient exhibits an obvious oscillation behavior, which is attributed to the multiple phonon scattering in the amorphous interlayer, and the oscillation period is further revealed to be consistent with the theoretical prediction by the two-beam interference equation. In addition, obvious phonon frequency shifts and phonon energy localization phenomena were observed in the amorphous interlayer. Finally, to improve phonon transmission, the interface morphology was further optimized via the annealing reconstruction technique, which results in re-crystallization of the amorphous interlayer and the decrease of ITR by ~21% as L=2 nm.

physics.app-ph↗
arXiv2023

Accelerating relaxation through Liouvillian exceptional point

We investigate speeding up of relaxation of Markovian open quantum systems with the Liouvillian exceptional point (LEP), where the slowest decay mode degenerate with a faster decay mode. The degeneracy significantly increases the gap of the Liouvillian operator, which determines the timescale of such systems in converging to stationarity, and hence accelerates the relaxation process. We explore an experimentally relevant three level atomic system, whose eigenmatrices and eigenspectra are obtained completely analytically. This allows us to gain insights in the LEP and examine respective dynamics with details. We illustrate that the gap can be further widened through Floquet engineering, which further accelerates the relaxation process. Finally, we extend this approach to analyze laser cooling of trapped ions, where vibrations (phonons) couple to the electronic states. An optimal cooling condition is obtained analytically, which agrees with both existing experiments and numerical simulations. Our study provides analytical insights in understanding LEP, as well as in controlling and optimizing dissipative dynamics of atoms and trapped ions.

quant-ph↗
arXiv2023

On Knowledge Editing in Federated Learning: Perspectives, Challenges, and Future Directions

As Federated Learning (FL) has gained increasing attention, it has become widely acknowledged that straightforwardly applying stochastic gradient descent (SGD) on the overall framework when learning over a sequence of tasks results in the phenomenon known as ``catastrophic forgetting''. Consequently, much FL research has centered on devising federated increasing learning methods to alleviate forgetting while augmenting knowledge. On the other hand, forgetting is not always detrimental. The selective amnesia, also known as federated unlearning, which entails the elimination of specific knowledge, can address privacy concerns and create additional ``space'' for acquiring new knowledge. However, there is a scarcity of extensive surveys that encompass recent advancements and provide a thorough examination of this issue. In this manuscript, we present an extensive survey on the topic of knowledge editing (augmentation/removal) in Federated Learning, with the goal of summarizing the state-of-the-art research and expanding the perspective for various domains. Initially, we introduce an integrated paradigm, referred to as Federated Editable Learning (FEL), by reevaluating the entire lifecycle of FL. Secondly, we provide a comprehensive overview of existing methods, evaluate their position within the proposed paradigm, and emphasize the current challenges they face. Lastly, we explore potential avenues for future research and identify unresolved issues.

cs.LG↗
arXiv2023

Dissecting Arbitrary-scale Super-resolution Capability from Pre-trained Diffusion Generative Models

Diffusion-based Generative Models (DGMs) have achieved unparalleled performance in synthesizing high-quality visual content, opening up the opportunity to improve image super-resolution (SR) tasks. Recent solutions for these tasks often train architecture-specific DGMs from scratch, or require iterative fine-tuning and distillation on pre-trained DGMs, both of which take considerable time and hardware investments. More seriously, since the DGMs are established with a discrete pre-defined upsampling scale, they cannot well match the emerging requirements of arbitrary-scale super-resolution (ASSR), where a unified model adapts to arbitrary upsampling scales, instead of preparing a series of distinct models for each case. These limitations beg an intriguing question: can we identify the ASSR capability of existing pre-trained DGMs without the need for distillation or fine-tuning? In this paper, we take a step towards resolving this matter by proposing Diff-SR, a first ASSR attempt based solely on pre-trained DGMs, without additional training efforts. It is motivated by an exciting finding that a simple methodology, which first injects a specific amount of noise into the low-resolution images before invoking a DGM's backward diffusion process, outperforms current leading solutions. The key insight is determining a suitable amount of noise to inject, i.e., small amounts lead to poor low-level fidelity, while over-large amounts degrade the high-level signature. Through a finely-grained theoretical analysis, we propose the Perceptual Recoverable Field (PRF), a metric that achieves the optimal trade-off between these two factors. Extensive experiments verify the effectiveness, flexibility, and adaptability of Diff-SR, demonstrating superior performance to state-of-the-art solutions under diverse ASSR environments.

cs.CV↗
arXiv2023

Energy coupling in intense laser solid interactions: material properties of gold

In the double-cone ignition inertial confinement fusion scheme, high density DT fuel is rapidly heated with high-flux fast electrons, which are generated by short and intense laser pulses. Gold cone target is usually used to shorten the distance between the critical surface and the compressed high density DT core. The material properties of solid gold may affect the generation and transport of fast electrons significantly, among which the effects of ionization and collision are the main concerns. In this work, the effects of ionization, collision and blow-off plasma on laser energy absorption rate are investigated using the LAPINS code: A three-stage model is adopted to explain the mechanism of fast electron generation and the change in laser energy absorption rate. With the increase of the charge state of Au ions, the laser-plasma interaction transfers to the later stage, resulting in a decrease in laser energy absorption rate. Collision has both beneficial and harmful effects. On one hand, collision provides a thermal pressure that makes it easier for electrons to escape into the potential well in front of the target and be accelerated in the second stage. On the other hand, collision increases stopping power and suppress electron recirculation within the target in the third stage. The vacuum sheath field behind the target enhances the electron circulation inside the target and thus improves the laser energy absorption, however this effect will be suppressed when the blow-off plasma density behind the target increases or collision is considered.

physics.plasm-ph↗
arXiv2023

Eeg2vec: Self-Supervised Electroencephalographic Representation Learning

Recently, many efforts have been made to explore how the brain processes speech using electroencephalographic (EEG) signals, where deep learning-based approaches were shown to be applicable in this field. In order to decode speech signals from EEG signals, linear networks, convolutional neural networks (CNN) and long short-term memory networks are often used in a supervised manner. Recording EEG-speech labeled data is rather time-consuming and laborious, while unlabeled EEG data is abundantly available. Whether self-supervised methods are helpful to learn EEG representation to boost the performance of EEG auditory-related tasks has not been well explored. In this work, we first propose a self-supervised model based on contrastive loss and reconstruction loss to learn EEG representations, and then use the obtained pre-trained model as a feature extractor for downstream tasks. Second, for two considered downstream tasks, we use CNNs and Transformer networks to learn local features and global features, respectively. Finally, the EEG data from other channels are mixed into the chosen EEG data for augmentation. The effectiveness of our method is verified on the EEG match-mismatch and EEG regression tasks of the ICASSP2023 Auditory EEG Challenge.

eess.AS↗
arXiv2023

Layer-adaptive Structured Pruning Guided by Latency

Structured pruning can simplify network architecture and improve inference speed. Combined with the underlying hardware and inference engine in which the final model is deployed, better results can be obtained by using latency collaborative loss function to guide network pruning together. Existing pruning methods that optimize latency have demonstrated leading performance, however, they often overlook the hardware features and connection in the network. To address this problem, we propose a global importance score SP-LAMP(Structured Pruning Layer-Adaptive Magnitude-based Pruning) by deriving a global importance score LAMP from unstructured pruning to structured pruning. In SP-LAMP, each layer includes a filter with an SP-LAMP score of 1, and the remaining filters are grouped. We utilize a group knapsack solver to maximize the SP-LAMP score under latency constraints. In addition, we improve the strategy of collect the latency to make it more accurate. In particular, for ResNet50/ResNet18 on ImageNet and CIFAR10, SP-LAMP is 1.28x/8.45x faster with +1.7%/-1.57% top-1 accuracy changed, respectively. Experimental results in ResNet56 on CIFAR10 demonstrate that our algorithm achieves lower latency compared to alternative approaches while ensuring accuracy and FLOPs.

cs.CV↗
arXiv2023

Semantic VAD: Low-Latency Voice Activity Detection for Speech Interaction

For speech interaction, voice activity detection (VAD) is often used as a front-end. However, traditional VAD algorithms usually need to wait for a continuous tail silence to reach a preset maximum duration before segmentation, resulting in a large latency that affects user experience. In this paper, we propose a novel semantic VAD for low-latency segmentation. Different from existing methods, a frame-level punctuation prediction task is added to the semantic VAD, and the artificial endpoint is included in the classification category in addition to the often-used speech presence and absence. To enhance the semantic information of the model, we also incorporate an automatic speech recognition (ASR) related semantic loss. Evaluations on an internal dataset show that the proposed method can reduce the average latency by 53.3% without significant deterioration of character error rate in the back-end ASR compared to the traditional VAD approach.

eess.AS↗
arXiv2023

CASA-ASR: Context-Aware Speaker-Attributed ASR

Recently, speaker-attributed automatic speech recognition (SA-ASR) has attracted a wide attention, which aims at answering the question ``who spoke what''. Different from modular systems, end-to-end (E2E) SA-ASR minimizes the speaker-dependent recognition errors directly and shows a promising applicability. In this paper, we propose a context-aware SA-ASR (CASA-ASR) model by enhancing the contextual modeling ability of E2E SA-ASR. Specifically, in CASA-ASR, a contextual text encoder is involved to aggregate the semantic information of the whole utterance, and a context-dependent scorer is employed to model the speaker discriminability by contrasting with speakers in the context. In addition, a two-pass decoding strategy is further proposed to fully leverage the contextual modeling ability resulting in a better recognition performance. Experimental results on AliMeeting corpus show that the proposed CASA-ASR model outperforms the original E2E SA-ASR system with a relative improvement of 11.76% in terms of speaker-dependent character error rate.

eess.AS↗
arXiv2023

VATLM: Visual-Audio-Text Pre-Training with Unified Masked Prediction for Speech Representation Learning

Although speech is a simple and effective way for humans to communicate with the outside world, a more realistic speech interaction contains multimodal information, e.g., vision, text. How to design a unified framework to integrate different modal information and leverage different resources (e.g., visual-audio pairs, audio-text pairs, unlabeled speech, and unlabeled text) to facilitate speech representation learning was not well explored. In this paper, we propose a unified cross-modal representation learning framework VATLM (Visual-Audio-Text Language Model). The proposed VATLM employs a unified backbone network to model the modality-independent information and utilizes three simple modality-dependent modules to preprocess visual, speech, and text inputs. In order to integrate these three modalities into one shared semantic space, VATLM is optimized with a masked prediction task of unified tokens, given by our proposed unified tokenizer. We evaluate the pre-trained VATLM on audio-visual related downstream tasks, including audio-visual speech recognition (AVSR), visual speech recognition (VSR) tasks. Results show that the proposed VATLM outperforms previous the state-of-the-art models, such as audio-visual pre-trained AV-HuBERT model, and analysis also demonstrates that VATLM is capable of aligning different modalities into the same space. To facilitate future research, we release the code and pre-trained models at https://aka.ms/vatlm.

eess.AS↗
arXiv2023

BASEN: Time-Domain Brain-Assisted Speech Enhancement Network with Convolutional Cross Attention in Multi-talker Conditions

Time-domain single-channel speech enhancement (SE) still remains challenging to extract the target speaker without any prior information on multi-talker conditions. It has been shown via auditory attention decoding that the brain activity of the listener contains the auditory information of the attended speaker. In this paper, we thus propose a novel time-domain brain-assisted SE network (BASEN) incorporating electroencephalography (EEG) signals recorded from the listener for extracting the target speaker from monaural speech mixtures. The proposed BASEN is based on the fully-convolutional time-domain audio separation network. In order to fully leverage the complementary information contained in the EEG signals, we further propose a convolutional multi-layer cross attention module to fuse the dual-branch features. Experimental results on a public dataset show that the proposed model outperforms the state-of-the-art method in several evaluation metrics. The reproducible code is available at https://github.com/jzhangU/Basen.git.

eess.AS↗
arXiv2023

Watermarking Text Generated by Black-Box Language Models

LLMs now exhibit human-like skills in various fields, leading to worries about misuse. Thus, detecting generated text is crucial. However, passive detection methods are stuck in domain specificity and limited adversarial robustness. To achieve reliable detection, a watermark-based method was proposed for white-box LLMs, allowing them to embed watermarks during text generation. The method involves randomly dividing the model vocabulary to obtain a special list and adjusting the probability distribution to promote the selection of words in the list. A detection algorithm aware of the list can identify the watermarked text. However, this method is not applicable in many real-world scenarios where only black-box language models are available. For instance, third-parties that develop API-based vertical applications cannot watermark text themselves because API providers only supply generated text and withhold probability distributions to shield their commercial interests. To allow third-parties to autonomously inject watermarks into generated text, we develop a watermarking framework for black-box language model usage scenarios. Specifically, we first define a binary encoding function to compute a random binary encoding corresponding to a word. The encodings computed for non-watermarked text conform to a Bernoulli distribution, wherein the probability of a word representing bit-1 being approximately 0.5. To inject a watermark, we alter the distribution by selectively replacing words representing bit-0 with context-based synonyms that represent bit-1. A statistical test is then used to identify the watermark. Experiments demonstrate the effectiveness of our method on both Chinese and English datasets. Furthermore, results under re-translation, polishing, word deletion, and synonym substitution attacks reveal that it is arduous to remove the watermark without compromising the original semantics.

cs.CL↗
arXiv2023

P4SGD: Programmable Switch Enhanced Model-Parallel Training on Generalized Linear Models on Distributed FPGAs

Generalized linear models (GLMs) are a widely utilized family of machine learning models in real-world applications. As data size increases, it is essential to perform efficient distributed training for these models. However, existing systems for distributed training have a high cost for communication and often use large batch sizes to balance computation and communication, which negatively affects convergence. Therefore, we argue for an efficient distributed GLM training system that strives to achieve linear scalability, while keeping batch size reasonably low. As a start, we propose P4SGD, a distributed heterogeneous training system that efficiently trains GLMs through model parallelism between distributed FPGAs and through forward-communication-backward pipeline parallelism within an FPGA. Moreover, we propose a light-weight, latency-centric in-switch aggregation protocol to minimize the latency of the AllReduce operation between distributed FPGAs, powered by a programmable switch. As such, to our knowledge, P4SGD is the first solution that achieves almost linear scalability between distributed accelerators through model parallelism. We implement P4SGD on eight Xilinx U280 FPGAs and a Tofino P4 switch. Our experiments show P4SGD converges up to 6.5X faster than the state-of-the-art GPU counterpar.

cs.DC↗
arXiv2023

Micromechanical measurements of local plastic events in granular materials

Understanding the relationship between micromechanics and macroscopic plastic deformation is vital for elucidating the deformation mechanism of amorphous solids, such as granular materials. In this study, we directly measure T1 events, which are topological rearrangements of particles, and the associated microscopic stresses in dense packings of photoelastic disks under pure shear. We observe a remarkable similarity between the evolution of the total number of T1 events and the global stress-strain curve. The competition between the birth and death of T1 events establishes a dynamic equilibrium after the yield strain, contributing to the overall plastic behavior. Despite the erratic stress fluctuations of individual T1 events, the local stresses decrease on average once T1 events occur, indicating their soft characteristics. Furthermore, we demonstrate that the microscopic stress fluctuations exhibit a long-range anisotropic spatial correlation similar to the Eschelby character, but with a distinct scaling of $r^{-1.5}$. Interestingly, we also find a striking similarity in the correlation functions of T1 and non-T1 regions. These findings establish a significant connection between macroscopic mechanical behavior and the elementary deformation of T1 events, shedding light on the fundamental understanding of granular materials as amorphous solids with contact-scale plastic deformation.

cond-mat.soft↗