检索结果-内蒙古大学图书馆

How graph neural networks learn: lessons from training dynamics 24

学校读者我要写书评

暂无评论

How graph neural networks learn: lessons from training dynam...

Proceedings of the 41st International Conference on Machine Learning

作者： Chenxiao Yang Qitian Wu David Wipf Ruoyu Sun Junchi Yan School of Artificial Intelligence & Department of Computer Science and Engineering & MoE Lab of AI Shanghai Jiao Tong University Amazon Web Services School of Data Science The Chinese University of Hong Kong Shenzhen and Shenzhen International Center for Industrial and Applied Mathematics Shenzhen Research Institute of Big Data

A long-standing goal in deep learning has been to characterize the learning behavior of black-box models in a more interpretable manner. For graph neural networks (GNNs), considerable advances have been made in formalizing what functions they can represent, but whether GNNs will learn desired functions during the optimization process remains less clear. To fill this gap, we study their training dynamics in function space. In particular, we find that the gradient descent optimization of GNNs implicitly leverages the graph structure to update the learned function, as can be quantified by a phenomenon which we call kernel-graph alignment. We provide theoretical explanations for the emergence of this phenomenon in the overparameterized regime and empirically validate it on real-world GNNs. This finding offers new interpretable insights into when and why the learned GNN functions generalize, highlighting their limitations in heterophilic graphs. Practically, we propose a parameter-free algorithm that directly uses a sparse matrix (i.e. graph adjacency) to update the learned function. We demonstrate that this embarrassingly simple approach can be as effective as GNNs while being orders-of-magnitude faster.

关键词：

Multi-Scale Accent Modeling and Disentangling for Multi-Speaker Multi-Accent Text-to-Speech Synthesis

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhou, Xuehao Zhang, Mingyang Zhou, Yi Wu, Zhizheng Li, Haizhou Department of Electrical and Computer Engineering National University of Singapore Singapore Hong Kong Generative AI Research and Development Centre The Hong Kong University of Science and Technology Hong Kong Shenzhen Research Institute of Big Data School of Data Science The Chinese University of Hong Kong Shenzhen China

Generating speech across different accents while preserving speaker identity is crucial for various real-world applications. However, accurately and independently modeling both speaker and accent characteristics in text-to-speech (TTS) systems is challenging due to the complex variations of accents and the inherent entanglement between speaker and accent identities. In this paper, we propose a novel approach for multispeaker multi-accent TTS synthesis that aims to synthesize speech for multiple speakers, each with various accents. Our approach employs a multi-scale accent modeling strategy to address accent variations on different levels. Specifically, we introduce both global (utterance level) and local (phoneme level) accent modeling to capture overall accent characteristics within an utterance and fine-grained accent variations across phonemes, respectively. To enable independent control of speakers and accents, we use the speaker embedding to represent speaker identity and achieve speaker-independent accent control through speaker disentanglement within the multi-scale accent modeling. Additionally, we present a local accent prediction model that enables our system to generate accented speech directly from phoneme inputs. We conduct extensive experiments on an English accented speech corpus. Experimental results demonstrate that our proposed system outperforms baseline systems in terms of speech quality and accent rendering for generating multispeaker multi-accent speech. Ablation studies further validate the effectiveness of different components in our proposed system. Copyright © 2024, The Authors. All rights reserved.

关键词： Quantum entanglement

Event Detection and Classification Using Machine Learning Applied to PMU data for the Western US Power System

学校读者我要写书评

暂无评论

Event Detection and Classification Using Machine Learning Ap...

Conference on Smart Grid Synchronized Measurements and analytics (SGSMA)

作者： Tianzhixi Yin Shaun S. Wulff John W. Pierre Brett G. Amidan AI & Data Analytics Pacific Northwest National Laboratory Richland WA USA Department of Mathematics and Statistics University of Wyoming Laramie WY USA Department of Electrical and Computer Engineering University of Wyoming Laramie WY USA Department of Mathematics Brigham Young University-Idaho Rexburg ID USA

ISBN: (数字)9798350312874

ISBN: (纸本)9798350312881

Smart grid technology enhances our comprehension and reliability of the power grid, leveraging Phasor Measurement Unit (PMU) data-time-synchronized, high-frequency measurements gathered across the US power grid. This paper employs machine learning techniques to effectively analyze the vast PMU data in Wide Area Monitoring Systems (WAMS) for power grid event detection and classification. Analyzing several months of real-world PMU data, the paper focuses on machine learning for fast, precise event detection and classification, corroborated by utility event logs. Practical challenges like feature extraction, dimensionality reduction, and model selection are addressed. A novel feature yielding improved results is discovered, and a supplementary algorithm for detecting small power grid faults is developed. The final algorithm is validated using a month-long real PMU data set, demonstrating its capability in accurately identifying power grid events in near real-time.

关键词： Machine learning algorithms Accuracy Event detection Time series analysis Feature extraction Phasor measurement units Classification algorithms

Untargeted Adversarial Attack on Knowledge Graph Embeddings

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Zhao, Tianzhe Chen, Jiaoyan Ru, Yanchi Lin, Qika Geng, Yuxia Liu, Jun School of Computer Science and Technology Xi'an Jiaotong University Xi’an China Department of Computer Science The University of Manchester Manchester United Kingdom National University of Singapore Singapore School of Computer Science Hangzhou Dianzi University Hangzhou China National Engineering Lab for Big Data Analytics Xi'an Jiaotong University Xi’an China

Knowledge graph embedding (KGE) methods have achieved great success in handling various knowledge graph (KG) downstream tasks. However, KGE methods may learn biased representations on low-quality KGs that are prevalent in the real world. Some recent studies propose adversarial attacks to investigate the vulnerabilities of KGE methods, but their attackers are target-oriented with the KGE method and the target triples to predict are given in advance, which lacks practicability. In this work, we explore untargeted attacks with the aim of reducing the global performances of KGE methods over a set of unknown test triples and conducting systematic analyses on KGE robustness. Considering logic rules can effectively summarize the global structure of a KG, we develop rule-based attack strategies to enhance the attack efficiency. In particular, we consider adversarial deletion which learns rules, applying the rules to score triple importance and delete important triples, and adversarial addition which corrupts the learned rules and applies them for negative triples as perturbations. Extensive experiments on two datasets over three representative classes of KGE methods demonstrate the effectiveness of our proposed untargeted attacks in diminishing the link prediction results. And we also find that different KGE methods exhibit different robustness to untargeted attacks. For example, the robustness of methods engaged with graph neural networks and logic rules depends on the density of the graph. But rule-based methods like NCRL are easily affected by adversarial addition attacks to capture negative rules. Copyright © 2024, The Authors. All rights reserved.

关键词： Graph embeddings

Prototype and Instance Contrastive Learning for Unsupervised Domain Adaptation in Speaker Verification

学校读者我要写书评

暂无评论

Prototype and Instance Contrastive Learning for Unsupervised...

International Symposium on Chinese Spoken Language Processing

作者： Wen Huang Bing Han Zhengyang Chen Shuai Wang Yanmin Qian Department of Computer Science and Engineering Auditory Cognition and Computational Acoustics Lab MoE Key Lab of Artificial Intelligence AI Institute Shanghai Jiao Tong University Shanghai Shenzhen Research Institute of Big Data The Chinese University of Hong Kong Shenzhen

ISBN: (数字)9798331516826

ISBN: (纸本)9798331516833

Speaker verification system trained on one domain usually suffers performance degradation when applied to another domain. To address this challenge, researchers commonly use feature distribution matching-based methods in unsupervised domain adaptation scenarios where some unlabeled target domain data is available. However, these methods often have limited performance improvement and lack generalization in various mismatch situations. In this paper, we propose Prototype and Instance Contrastive Learning (PICL), a novel method for unsupervised domain adaptation in speaker verification through dual-level contrastive learning. For prototype contrastive learning, we generate pseudo labels via clustering to create dynamically updated prototype representations, aligning instances with their corresponding class or cluster prototypes. For instance contrastive learning, we minimize the distance between different views or augmentations of the same instance, ensuring robust and invariant representations resilient to variations like noise. This dual-level approach provides both high-level and low-level supervision, leading to improved generalization and robustness of the speaker verification model. Unlike previous studies that only evaluated mismatches in one situation, we have conducted relevant explorations on various datasets and achieved state-of-the-art performance currently, which also proves the generalization of our method.

关键词： Degradation Adaptation models Noise Prototypes Contrastive learning Robustness

Attention-Constrained Inference For Robust Decoder-Only Text-to-Speech

学校读者我要写书评

暂无评论

Attention-Constrained Inference For Robust Decoder-Only Text...

IEEE Spoken Language Technology Workshop

作者： Hankun Wang Chenpeng Du Yiwei Guo Shuai Wang Xie Chen Kai Yu Department of Computer Science and Engineering MoE Key Lab of Artificial Intelligence AI Institute X-LANCE Lab Shanghai Jiao Tong University Shanghai China Shenzhen Research Institute of Big Data The Chinese University of Hong Kong Shenzhen China

ISBN: (数字)9798350392258

ISBN: (纸本)9798350392265

Recent popular decoder-only text-to-speech models are known for their ability of generating natural-sounding speech. However, such models sometimes suffer from word skipping and repeating due to the lack of explicit monotonic alignment constraints. In this paper, we notice from the attention maps that some particular attention heads of the decoder-only model indicate the alignments between speech and text. We call the attention maps of those heads Alignment-Emerged Attention Maps (AEAMs). Based on this discovery, we propose a novel inference method without altering the training process, named Attention-Constrained Inference (ACI), to facilitate monotonic synthesis. It first identifies AEAMs using the Attention Sweeping algorithm and then applies constraining masks on AEAMs. Our experimental results on decoderonly TTS model VALL-E show that the WER of synthesized speech is reduced by up to 20.5% relatively with ACI while the naturalness and speaker similarity are comparable.

关键词： Training Conferences Robustness Inference algorithms Text to speech Optimization

Attention-based Encoder-Decoder Network for End-to-End Neural Speaker Diarization with Target Speaker Attractor

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Chen, Zhengyang Han, Bing Wang, Shuai Qian, Yanmin MoE Key Lab of Artificial Intelligence AI Institute X-LANCE Lab Department of Computer Science and Engineering Shanghai Jiao Tong University China Shenzhen Research Institute of Big Data Shenzhen China

This paper proposes a novel Attention-based Encoder-Decoder network for End-to-End Neural speaker Diarization (AED-EEND). In AED-EEND system, we incorporate the target speaker enrollment information used in target speaker voice activity detection (TS-VAD) to calculate the attractor, which can mitigate the speaker permutation problem and facilitate easier model convergence. In the training process, we propose a teacher-forcing strategy to obtain the enrollment information using the ground-truth label. Furthermore, we propose three heuristic decoding methods to identify the enrollment area for each speaker during the evaluation process. Additionally, we enhance the attractor calculation network LSTM used in the end-to-end encoder-decoder based attractor calculation (EEND-EDA) system by incorporating an attention-based model. By utilizing such an attention-based attractor decoder, our proposed AED-EEND system outperforms both the EEND-EDA and TS-VAD systems with only 0.5s of enrollment data. © 2023, CC BY-NC-SA.

关键词： Heuristic methods

Optimizing Ethanol Production in Escherichia Coli Using a Hybrid of Particle Swarm Optimization and Artificial Bee Colony 22

学校读者我要写书评

暂无评论

Optimizing Ethanol Production in Escherichia Coli Using a Hy...

Proceedings of the 6th International Conference on Advances in Artificial Intelligence

作者： Mohamad Faiz Dzulkalnine Mohd Saberi Mohamad Yee Wen Choon Muhammad Akmal Remli Hany Alashwal Artificial Intelligence and Bioinformatics Research Group Faculty of Computing Universiti Teknologi Malaysia Malaysia Health Data Science Lab Department of Genetics and Genomics College of Medical and Health Sciences United Arab Emirates University UAE and Big Data Analytics Center United Arab Emirates University UAE Institute for Artificial Intelligence and Big Data Universiti Malaysia Kelantan Malaysia and Department of Data Science Universiti Malaysia Kelantan Malaysia Department of Computer Science and Software Engineering College of Information Technology United Arab Emirates University UAE and Big Data Analytics Center United Arab Emirates University UAE

ISBN: (纸本)9781450396943

Metabolic engineering for biomass production using microorganisms’ cell has received considerable attention in recent years. This is due to the biomass products being extensively used in the field of food additives, supplements, pharmaceuticals, and polymer materials. In this paper, ethanol production in Escherichia coli (E. coli) is the desired product. Sugarcane and corn are often used to produce ethanol. However, one of the problems to produce adequate amounts of ethanol is that large areas are needed to plant sugarcane and corn. Furthermore, the amount of time for the process of dry milling and wet milling is high, which are 40 to 50 hours and 24 to 48 hours, respectively. The wet laboratory is also having limitation on the production of ethanol in microorganisms because the amount of the ethanol produced is not satisfying. Hence, a lot of metabolic engineering techniques is introduced to enhance the production of ethanol in E. coli, such as gene knockout strategy, but the production is yet to meet the demand. Therefore, this paper proposes a hybrid algorithm of Particle Swarm Optimization with the Artificial Bee Colony algorithm (PSOABC) to identify the optimal set of gene knockout strategy to improve the ethanol production in E. coli. A list of genes to knockout, production of the desired product, and growth rate are presented in this paper. PSOABC has shown better performance in terms of production, growth rate and accuracy.

关键词： Gene Knockouts Particle Swarm Optimization Artificial Bee Colony Ethanol Flux Balance Analysis

ATTENTION-CONSTRaiNED INFERENCE FOR ROBUST DECODER-ONLY TEXT-TO-SPEECH

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Wang, Hankun Du, Chenpeng Guo, Yiwei Wang, Shuai Chen, Xie Yu, Kai MoE Key Lab of Artificial Intelligence AI Institute X-LANCE Lab Department of Computer Science and Engineering Shanghai Jiao Tong University Shanghai China Shenzhen Research Institute of Big Data The Chinese University of Hong Kong Shenzhen China

Recent popular decoder-only text-to-speech models are known for their ability of generating natural-sounding speech. However, such models sometimes suffer from word skipping and repeating due to the lack of explicit monotonic alignment constraints. In this paper, we notice from the attention maps that some particular attention heads of the decoder-only model indicate the alignments between speech and text. We call the attention maps of those heads Alignment-Emerged Attention Maps (AEAMs). Based on this discovery, we propose a novel inference method without altering the training process, named Attention-Constrained Inference (ACI), to facilitate monotonic synthesis. It first identifies AEAMs using the Attention Sweeping algorithm and then applies constraining masks on AEAMs. Our experimental results on decoder-only TTS model VALL-E show that the WER of synthesized speech is reduced by up to 20.5% relatively with ACI while the naturalness and speaker similarity are comparable. Copyright © 2024, The Authors. All rights reserved.

关键词： Decoding