检索结果-内蒙古大学图书馆

Deep learning-based image deconstruction method with maintained saliency

NEURAL NETWORKS 2022年 155卷 224-241页

作者： Fujimoto, Keisuke Hayashi, Kojiro Katayama, Risa Lee, Sehyung Liang, Zhen Yoshida, Wako Ishii, Shin Kyoto Univ Grad Sch Informat Kyoto 6068501 Japan ATR Neural Informat Anal Labs Kyoto 6190288 Japan Shenzhen Univ Hlth Sci Ctr Sch Biomed Engn Shenzhen Peoples R China Univ Oxford Nuffield Dept Clin Neurosci Oxford England 36-1 Yoshida Honmachi Sakyo Ku Kyoto 6068501 Japan

Visual properties that primarily attract bottom-up attention are collectively referred to as saliency. In this study, to understand the neural activity involved in top-down and bottom-up visual attention, we aim to prepare pairs of natural and unnatural images with common saliency. For this purpose, we propose an image transformation method based on deep neural networks that can generate new images while maintaining the consistent feature map, in particular the saliency map. This is an ill -posed problem because the transformation from an image to its corresponding feature map could be many-to-one, and in our particular case, the various images would share the same saliency map. Although stochastic image generation has the potential to solve such ill-posed problems, the most existing methods focus on adding diversity of the overall style/touch information while maintaining the naturalness of the generated images. To this end, we developed a new image transformation method that incorporates higher-dimensional latent variables so that the generated images appear unnatural with less context information but retain a high diversity of local image structures. Although such high-dimensional latent spaces are prone to collapse, we proposed a new regularization based on Kullback-Leibler divergence to avoid collapsing the latent distribution. We also conducted human experiments using our newly prepared natural and corresponding unnatural images to measure overt eye movements and functional magnetic resonance imaging, and found that those images induced distinctive neural activities related to top-down and bottom-up attentional processing. (C) 2022 The Author(s). Published by Elsevier Ltd.

关键词： Attention Image transformation Saliency map Deep learning variational autoencoder Functional magnetic resonance imaging

来源：评论

学校读者我要写书评

暂无评论

A Gaussian RBM with binary auxiliary units

引用

INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS 2022年第9期13卷 2425-2433页

作者： Zhang, Jian Ding, Shifei Sun, Tongfeng Guo, Lili China Univ Min & Technol Sch Comp Sci & Technol Xuzhou 221116 Jiangsu Peoples R China Minist Educ Peoples Republ China Mine Digitizat Engn Res Ctr Xuzhou 221116 Jiangsu Peoples R China

Restricted Boltzmann Machines (RBM) have been widely applied in image processing. For RBM-based models on image recognition and image generation tasks, extracting expressive real-valued features and alleviating the overfitting problem are extremely important. In this paper, we propose a Gaussian Restricted Boltzmann Machine with binary Auxiliary units (GARBM), which designs binary auxiliary units in its visible layer and constructs parameterized real-valued features in its hidden layer. Specifically, based on the designed energy function in GARBM, activated auxiliary units are directly used to control probabilities of visible units and hidden units to extract real-valued features. Moreover, auxiliary units and their resulting feature selection mechanism not only alleviate the "gradient-variance" problem, but also provide certain randomness to other units to alleviate overfitting without introducing more hyperparameters. To build more effective deep models, we propose GARBM-based deep neural networks, and the effectiveness of proposed neural networks is verified in experiments.

关键词： Restricted Boltzmann machine Truncated Gaussian distribution Deep belief net variational autoencoder Auxiliary unit

来源：评论

学校读者我要写书评

暂无评论

EfficientFi: Toward Large-Scale Lightweight WiFi Sensing via CSI Compression

引用

IEEE INTERNET OF THINGS JOURNAL 2022年第15期9卷 13086-13095页

作者： Yang, Jianfei Chen, Xinyan Zou, Han Wang, Dazhuo Xu, Qianwen Xie, Lihua Nanyang Technol Univ Sch Elect & Elect Engn Singapore Singapore Univ Calif Berkeley Dept Elect Engn & Comp Sci Berkeley CA 94720 USA KTH Royal Inst Technol Dept Elect Power & Energy Syst S-10044 Stockholm Sweden

WiFi technology has been applied to various places due to the increasing requirement of high-speed Internet access. Recently, besides network services, WiFi sensing is appealing in smart homes since it is device free, cost effective and privacy preserving. Though numerous WiFi sensing methods have been developed, most of them only consider single smart home scenario. Without the connection of powerful cloud server and massive users, large-scale WiFi sensing is still difficult. In this article, we first analyze and summarize these obstacles, and propose an efficient large-scale WiFi sensing framework, namely, EfficientFi. The EfficientFi works with edge computing at WiFi access points and cloud computing at center servers. It consists of a novel deep neural network that can compress fine-grained WiFi channel state information (CSI) at edge, restore CSI at cloud, and perform sensing tasks simultaneously. A quantized autoencoder and a joint classifier are designed to achieve these goals in an end-to-end fashion. To the best of our knowledge, the EfficientFi is the first Internet of Things-cloud-enabled WiFi sensing framework that significantly reduces communication overhead while realizing sensing tasks accurately. We utilized human activity recognition (HAR) and identification via WiFi sensing as two case studies, and conduct extensive experiments to evaluate the EfficientFi. The results show that it compresses CSI data from 1.368 Mb/s to 0.768 kb/s with extremely low error of data reconstruction and achieves over 98% accuracy for HAR.

关键词： Sensors Wireless fidelity Servers Cloud computing Feature extraction Internet of Things Deep learning Channel state information (CSI) deep neural network discrete representation learning multitask learning variational autoencoder WiFi-based sensing

来源：评论

学校读者我要写书评

暂无评论

WakaVT: A Sequential variational Transformer for Waka Generation

引用

NEURAL PROCESSING LETTERS 2022年第2期54卷 731-750页

作者： Takeishi, Yuka Niu, Mingxuan Luo, Jing Jin, Zhong Yang, Xinyu Xi An Jiao Tong Univ Sch Foreign Studies 28 Xianning West Rd Xian 710049 Shaanxi Peoples R China Xi An Jiao Tong Univ Sch Comp Sci & Technol 28 Xianning West Rd Xian 710049 Shaanxi Peoples R China

Poetry generation has long been a challenge for artificial intelligence. In the scope of Japanese poetry generation, many researchers have paid attention to Haiku generation, but few have focused on Waka generation. To further explore the creative potential of natural language generation systems in Japanese poetry creation, we propose a novel Waka generation model, WakaVT, which automatically produces Waka poems given user-specified keywords. Firstly, an additive mask-based approach is presented to satisfy the form constraint. Secondly, the structures of Transformer and variational autoencoder are integrated to enhance the quality of generated content. Specifically, to obtain novelty and diversity, WakaVT employs a sequence of latent variables, which effectively captures word-level variability in Waka data. To improve linguistic quality in terms of fluency, coherence, and meaningfulness, we further propose the fused multilevel self-attention mechanism, which properly models the hierarchical linguistic structure of Waka. To the best of our knowledge, we are the first to investigate Waka generation with models based on Transformer and/or variational autoencoder. Both objective and subjective evaluation results demonstrate that our model outperforms baselines significantly.

关键词： Waka generation Self-attention mechanism variational autoencoder Linguistic quality

来源：评论

学校读者我要写书评

暂无评论

Synthetic PPG Signal Generation to Improve Coronary Artery Disease Classification: Study With Physical Model of Cardiovascular System

引用

IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS 2022年第5期26卷 2136-2146页

作者： Mazumder, Oishee Banerjee, Rohan Roy, Dibyendu Bhattacharya, Sakyajit Ghose, Avik Sinha, Aniruddha Tata Consultancy Serv Ltd TCS Res Kolkata 700156 W Bengal India

This paper presents a novel approach of generating synthetic Photoplethysmogram (PPG) data using a physical model of the cardiovascular system to improve classifier performance with a combination of synthetic and real data. The physical model is an in-silico cardiac computational model, consisting of a four-chambered heart with electrophysiology, hemodynamic, and blood pressure auto-regulation functionality. Starting with a small number of measured PPG data, the cardiac model is used to synthesize healthy as well as PPG time-series pertaining to coronary artery disease (CAD) by varying pathophysiological parameters. A variational autoencoder (VAE) structure is proposed to derive a statistical feature space for CAD classification. Results are presented in two perspectives namely, (i) using artificially reduced real disease data and (ii) using all the real disease data. In both cases, by augmenting with the synthetic data for training, the performance (sensitivity, specificity) of the classifier changes from (i) (0.65, 1) to (1, 0.9) and (ii) (1, 0.95) to (1, 1). The proposed hybrid approach of combining physical modelling and statistical feature space selection generates realistic PPG data with pathophysiological interpretation and can outperform a baseline Generative Adversarial Network (GAN) architecture with a relatively small amount of real data for training. This proposed method could aid as a substitution technique for handling the problem of bulk data required for training machine learning algorithms for cardiac health-care applications.

关键词： Solid modeling Electrocardiography Diseases Data models Training Hemodynamics Bioinformatics Cardiac computational model coronary artery disease photoplethysmogram synthetic data variational autoencoder

来源：评论

学校读者我要写书评

暂无评论

Situation-Aware Multivariate Time Series Anomaly Detection Through Active Learning and Contrast VAE-Based Models in Large Distributed Systems

引用

IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS 2022年第9期40卷 2746-2765页

作者： Li, Zhihan Zhao, Youjian Geng, Yitong Zhao, Zhanxiang Wang, Hanzhang Chen, Wenxiao Jiang, Huai Vaidya, Amber Su, Liangfei Pei, Dan Tsinghua Univ Dept Comp Sci & Technol Beijing 100084 Peoples R China eBay Inc San Jose CA 95125 USA

The massive amounts of monitoring data in network applications bring an urgent need for intelligent operation in large distributed systems. The key problem is precisely detecting anomalies in multivariate time series (MTS) monitoring metrics with the awareness of different application scenarios. Unsupervised NITS anomaly detection methods aim at detecting data anomalies from historical MTS without considering the out-of-band information (including user feedback and background information like code deployment status), which leads to poor performance in practice. To take advantage of the out-of-band information, we propose ACVAE, an MTS anomaly detection algorithm through active learning and contrast VAE-based detection models, which simultaneously learns NITS data's normal and anomalous patterns for anomaly detection. We also use a learnable prior to capture system status from the background information. Moreover, we propose a query model for VAE-based methods, which can learn to query labels of the must useful instances to train the detection model. We evaluate our algorithm on three different monitoring situations in eBay's search back-end systems. ACVAE achieves a range F1 score of 0.68 similar to 0.96 with only 3% labels, significantly outperforming the best competing methods by 0.18 similar to 0.50, and even better than a supervised ensemble method designed by domain experts in eBay.

关键词： Anomaly detection out-of-band information multivariate time series active learning variational autoencoder

来源：评论

学校读者我要写书评

暂无评论

Non-episodic variational Weight Imprinting for Few-Shot Learning

引用

SN Computer Science 2023年第3期4卷 303页

作者： Padma, V. Bhagvati, Chakravarthy School of Computer and Information Sciences University of Hyderabad Hyderabad India

Few-shot learning methods enable models to recognize novel classes from one or only a few examples. Most of the existing methods use episodic training which require multiple training models and/or variable memory requirements. We propose a non-episodic model, which combines a β-variational auto encoder (β-VAE) with the cosine similarity classifier (CSC). It is trained only once with fixed memory requirement and can be used for all test settings. We have also created our own printed character dataset with style attributes that includes six Indic scripts to analyze the effect of β-VAE’s disentangled content on few-shot classification accuracy. In our model, the encoder of β-VAE generates both the content and style distribution where only the content is used by the CSC. The network is trained in an end-to-end fashion which enables the encoder to disentangle the content features needed for classification from style attributes. After the training phase, the weights of novel classes are generated by normalizing the encoder’s content representation and used for testing. In case of Omniglot, we got 98.91% classification accuracy for 5-way 1-shot which is on par/better than the state-of-the art episodic methods. Our ablation study shows that β-VAE with CSC network outperforms CSC network on Indic script dataset. Our proposed model which is trained only once with fixed memory requirement gives comparable results with episodic and non-episodic methods as the encoder feeds the disentangled content features to the classifier. In a few settings, it also outperforms the episodic methods. © 2023, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.

关键词： Cosine similarity classifier Few-shot learning One shot learning variational autoencoder

来源：评论

学校读者我要写书评

暂无评论

Efficient and Accurate Object Extraction from Scanned Maps by Leveraging External Data and Learning Representative Context 23

Efficient and Accurate Object Extraction from Scanned Maps b...

引用

31st ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (ACM SIGSPATIAL GIS)

作者： Duan, Weiwei Univ Southern Calif Los Angeles CA 90089 USA

ISBN: (纸本)9798400701689

This paper introduces innovative approaches for extracting geographic objects from scanned map images. Overcoming the challenges associated with labor-intensive data labeling and inaccurate extraction results, we present two data labeling methods and an accurate-enhancing extraction method. The experiment in this paper shows that our approaches outperform the baselines.

关键词： Generative models variational autoencoder Transformer Object Detection Map Processing Geospatial Information Sciences

来源：评论

学校读者我要写书评

暂无评论

Causal Disentangled variational Auto-Encoder for Preference Understanding in Recommendation 23

Causal Disentangled Variational Auto-Encoder for Preference ...

引用

46th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR)

作者： Wang, Siyu Chen, Xiaocong Sheng, Quan Z. Zhang, Yihong Yao, Lina Univ New South Wales Sydney NSW Australia Macquarie Univ Sydney NSW Australia Osaka Univ Osaka Japan CSIRO Data61 Eveleigh Australia

ISBN: (纸本)9781450394086

Recommendation models are typically trained on observational user interaction data, but the interactions between latent factors in users' decision-making processes lead to complex and entangled data. Disentangling these latent factors to uncover their underlying representation can improve the robustness, interpretability, and controllability of recommendation models. This paper introduces the Causal Disentangled variational Auto-Encoder (CaD-VAE), a novel approach for learning causal disentangled representations from interaction data in recommender systems. The CaD-VAE method considers the causal relationships between semantically related factors in real-world recommendation scenarios, rather than enforcing independence as in existing disentanglement methods. The approach utilizes structural causal models to generate causal representations that describe the causal relationship between latent factors. The results demonstrate that CaD-VAE outperforms existing methods, offering a promising solution for disentangling complex user behavior data in recommendation systems.

关键词： Recommender Systems Causal Disentangled Representation variational autoencoder

来源：评论

学校读者我要写书评

暂无评论

LyricJam Sonic: A Generative System for Real-Time Composition and Musical Improvisation 12th

LyricJam Sonic: A Generative System for Real-Time Compositio...

引用

12th International Conference on Artificial Intelligence in Music, Sound, Art and Design (EvoMUSART) Held as Part of EvoStar Conference

作者： Vechtomova, Olga Sahu, Gaurav Univ Waterloo Waterloo ON Canada

ISBN: (纸本)9783031299551;9783031299568

Electronic music artists and sound designers have unique workflow practices that necessitate specialized approaches for developing music information retrieval and creativity support tools. Furthermore, electronic music instruments, such as modular synthesizers, have near-infinite possibilities for sound creation and can be combined to create unique and complex audio paths. The process of discovering interesting sounds is often serendipitous and impossible to replicate. For this reason, many musicians in electronic genres record audio output at all times while they work in the studio. Subsequently, it is difficult for artists to rediscover audio segments that might be suitable for use in their compositions from thousands of hours of recordings. In this paper, we describe LyricJam Sonic, a creative tool for musicians to rediscover their previous recordings, re-contextualize them with other recordings, and create original live music compositions in real-time. A bi-modal AI-driven approach uses generated lyric lines to find compatible audio clips from the artist's past studio recordings, and uses them to generate new lyric lines, which in turn are used to find other clips, thus creating a continuous and evolving stream of music and lyrics. The intent is to keep the artists in a state of creative flow conducive to music creation rather than taking them into an analytical/critical state of deliberately searching for past audio segments. The system can run in either a fully autonomous mode without user input, or in a live performance mode, where the artist plays live music, while the system "listens" and creates a continuous stream of music and lyrics in response (LyricJam Sonic: https://*** Demo videos: https://***/view/supplementarymaterial-for-evo/home).

关键词： Generative Music Lyrics Generation Neural Network variational autoencoder Generative Adversarial Network

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：