检索结果-内蒙古大学图书馆

24th IEEE International Conference on data Mining Workshops, ICDMW 2024

作者： Fuchs, Marco Catholic University of Eichstätt-Ingolstadt Mathematical Institute for Machine Learning and Data Science Ingolstadt Germany

ISBN: (纸本)9798331530631

We introduce a method for partitioning a time series into segments. The method extends the recently introduced Fast Low-Cost Semantic Segmentation (FLUSS) algorithm to increase its robustness against noise and to automatically learn two required hyperparameters. FLUSS is part of the Matrix Profile framework and conceptualises the relationship between every subsequence and its nearest-neighbor as an arc spanning over the time series. It then determines segment boundaries by selecting time indexes which are spanned over by significantly fewer arcs than an average index, such that most subsequences on either side of the time index find their closest neighbor at the same side of the index. We extend FLUSS in two directions. First, we consider the Z nearest neighbors of each subsequence to compute its neighbors' concentration along the time series as a measure for how characteristic a sequence itself is for a particular segment. This measure is then used to filter out non-characteristic sequences, aimed at increasing the robustness of the segmentation against noise. Second, FLUSS requires the user to set two important yet hard to determine hyperparameters, namely the number of segments and the length of exclusion zones around the index of an already found segment boundary. We introduce a procedure to learn both hyperparameters implicitly by setting the tolerance parameter of an iterative end-point fit algorithm, which extracts (local) minima from a meta time series denoting the number of arcs that cross a time series index. We test the method on a set of 32 diverse and publicly available univariate time series, where it demonstrates promising results. © 2024 IEEE.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

D’OH: Decoder-Only Random Hypernetworks for Implicit Neural Representations 17th

D’OH: Decoder-Only Random Hypernetworks for Implicit Neura...

引用

17th Asian Conference on Computer Vision, ACCV 2024

作者： Gordon, Cameron E. MacDonald, Lachlan Saratchandran, Hemanth Lucey, Simon Australian Institute for Machine Learning University of Adelaide AdelaideSA5000 Australia Mathematical Institute for Data Science John Hopkins University BaltimoreMD21218 United States

ISBN: (纸本)9789819609628

Deep implicit functions have been found to be an effective tool for efficiently encoding all manner of natural signals. Their attractiveness stems from their ability to compactly represent signals with little to no offline training data. Instead, they leverage the implicit bias of deep networks to decouple hidden redundancies within the signal. In this paper, we explore the hypothesis that additional compression can be achieved by leveraging redundancies that exist between layers. We propose to use a novel runtime decoder-only hypernetwork – that uses no offline training data – to better exploit cross-layer parameter redundancy. Previous applications of hypernetworks with deep implicit functions have employed feed-forward encoder/decoder frameworks that rely on large offline datasets that do not generalize beyond the signals they were trained on. We instead present a strategy for the optimization of runtime deep implicit functions for single-instance signals through a Decoder-Only randomly projected Hypernetwork (D’OH). By directly changing the latent code dimension, we provide a natural way to vary the memory footprint of neural representations without the costly need for neural architecture search on a space of alternative low-rate structures. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Decoding

来源：评论

学校读者我要写书评

暂无评论

Hybrid speech enhancement in modulation domain

引用

Multimedia Tools and Applications 2024年 1-31页

作者： P S, Praveen Kumar H S, Jayanna Technical Lead-Machine Learning Merit Data and Technology Tamil Nadu Chennai India Department of Information Science and Engineering Siddaganga Institute of Technology Karnataka Tumkur India

This paper presents a comprehensive study on speech enhancement (SE) techniques, particularly focusing on the utilization of the discrete cosine transform (DCT) in the modulation domain (MD) in combination with the minimum mean square error (MMSE) and Kalman filter (KF) algorithms. The research investigates the efficacy of these techniques in improving speech quality and intelligibility under various noisy conditions, such as babble, drilling, horn, train, and traffic noises, across different signal-to-noise ratio (SNR) levels. Additionally, the study evaluates the performance of recurrent neural network (RNN) algorithms alongside traditional SE approaches. The study highlights the advantages of employing DCT in the modulation domain over the traditional discrete Fourier transform (DFT) in SE applications. Experimental results demonstrate significant improvements in objective quality metrics, including Perceptual Evaluation of Speech Quality (PESQ), composite measures, and short-time objective intelligibility (STOI), when implementing SE algorithms using discrete cosine transform in modulation domain. The proposed hybrid speech enhancement techniques, leveraging minimum mean square error and Kalman filter in combination with discrete cosine transform in modulation domain, outperform individual speech enhancement algorithms, showcasing superior noise reduction capabilities. This paper contributes to the advancement of speech enhancement methodologies, particularly in real-world noisy environments, and underscores the effectiveness of discrete cosine transform-based approaches in enhancing speech quality and intelligibility. © The Author(s), under exclusive licence to Springer science+Business Media, LLC, part of Springer Nature 2024.

关键词： Recurrent neural networks

来源：评论

学校读者我要写书评

暂无评论

Impact of Immersive Technology on Physiological Parameters in Psychiatric Disorders 13

Impact of Immersive Technology on Physiological Parameters i...

引用

13th International Conference on System Modeling and Advancement in Research Trends, SMART 2024

作者： Kataria, Anjali Singh, Gurjinder Sandhu, Jasminder Chitkara Institute of Engineering and Technology Punjab India Iilm University Department of Machine Learning and Data Science Noida India

ISBN: (纸本)9798350380583

The brain is an essential component that regulates the general functioning of the body. The brain consists of millions of neurons that govern human behavior in response to sensory stimuli. To understand the mental functioning of the brain, it is important to assess its neural signals. Presently, the exponential expansion of prescription medications and substances, including alcohol and drugs, is exerting a significant influence on human psychology. The rising prevalence of different health disorders can be attributed to the increasing number of cases. The sensorimotor cortex, which in humans is in charge of coordinating both physical and cognitive activities, can be used to visualize brain signals. The present research describes an empirical investigation that employs Virtual Reality therapy to examine the affective condition of individuals suffering from drug addiction. The study utilized the Emotiv Insight 5-channel EEG headset to collect EEG data from 50 participants, and the Emotiv Pro software was employed for data analysis. The study findings indicate that Virtual Reality treatment has a substantial positive impact on the stress levels and concentration of persons suffering from drug addiction. © 2024 IEEE.

关键词： Electroencephalography

来源：评论

学校读者我要写书评

暂无评论

LSTM based Soft-Sensor for Estimating Nitrate Concentration in Aquaponics pond 3

LSTM based Soft-Sensor for Estimating Nitrate Concentration ...

引用

3rd International Conference for Innovation in Technology, INOCON 2024

作者： Dharshan, A. Kumar, Purushottam Ravimaran, S. Srinivasulu Reddy, U. Saranathan College of Engineering Department of Artificial Intelligence and Data Science Trichy India National Institute of Technology Artificial Intelligence Machine Learning & Data Analytics Lab Trichy India National Institute of Technology Department of Computer Applications Machine Learning & Data Analytics Lab Trichy India

ISBN: (纸本)9798350381931

In the field of aquaponics, where fish and plants coexist in a symbiotic environment, closely monitoring nitrate levels in the water is crucial due to their profound impact on aquatic and plant well-being. Traditional nitrate measurement methods are often time-consuming and costly. Various approaches, including first principles, IoT-based sensors, and machine learning-based soft sensors, have been attempted to address this challenge. However, these efforts face challenges such as expensive sensors, infrequent data collection, multistage data processing using limited sensor types, and the need for regular maintenance like cleaning and calibration. Additionally, varied environmental conditions affect sensor suitability for different water environments, and even some machine learning-based soft sensors have proven inaccurate. In response, soft sensors, especially deep learning-based ones, have gained prominence in industrial applications for their adaptability and accuracy. These sensors provide real-time insights into complex processes without requiring expensive hardware. In this study, an innovative solution was introduced using Long Short-Term Memory (LSTM) technology, a neural network architecture in deep learning known for capturing complex temporal patterns. LSTM is well-suited for modeling and predicting nitrate concentration changes in aquaponics, trained with extensive data collected from various aquaponic ponds. Through rigorous evaluation, a remarkable MSE value of 0.00074 and an impressive R-squared score of 0.98 were achieved, holding potential for scaling up to commercial applications, benefiting aquaponics operations, supporting researchers, and enhancing sustainability and productivity in aquaponic systems. © 2024 IEEE.

关键词： Nitrates

来源：评论

学校读者我要写书评

暂无评论

Improving Generalization and Convergence by Enhancing Implicit Regularization 38

Improving Generalization and Convergence by Enhancing Implic...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Wang, Mingze Wang, Jinbo He, Haotian Wang, Zilin Huang, Guanhua Xiong, Feiyu Li, Zhiyu Weinan, E. Wu, Lei School of Mathematical Sciences Peking University China Center for Machine Learning Research Peking University China China AI for Science Institute China School of Data Science University of Science and Technology of China China ByteDance Research China

In this work, we propose an Implicit Regularization Enhancement (IRE) framework to accelerate the discovery of flat solutions in deep learning, thereby improving generalization and convergence. Specifically, IRE decouples the dynamics of flat and sharp directions, which boosts the sharpness reduction along flat directions while maintaining the training stability in sharp directions. We show that IRE can be practically incorporated with generic base optimizers without introducing significant computational overload. Experiments show that IRE consistently improves the generalization performance for image classification tasks across a variety of benchmark datasets (CIFAR-10/100, ImageNet) and models (ResNets and ViTs). Surprisingly, IRE also achieves a 2× speed-up compared to AdamW in the pre-training of Llama models (of sizes ranging from 60M to 229M) on datasets including Wikitext-103, Minipile, and Openwebtext. Moreover, we provide theoretical guarantees, showing that IRE can substantially accelerate the convergence towards flat minima in sharpness-aware minimization (SAM). © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Optimization of Random Feature Method in the High-Precision Regime

引用

Communications on Applied Mathematics and Computation 2024年第2期6卷 1490-1517页

作者： Jingrun Chen Weinan E Yifei Sun School of Mathematical Sciences and Suzhou Institute for Advanced Research Suzhou 215006JiangsuChina University of Science and Technology of China Hefei 230026AnhuiChina Center for Machine Learning Research and School of Mathematical Sciences Peking UniversityBeijing 100871China AI for Science Institute Beijing 100084China School of Mathematical Sciences Soochow UniversitySuzhou 215006JiangsuChina

machine learning has been widely used for solving partial differential equations(PDEs)in recent years,among which the random feature method(RFM)exhibits spectral accuracy and can compete with traditional solvers in terms of both accuracy and ***,the optimization problem in the RFM is more difficult to solve than those that arise in traditional *** the broader machine-learning research,which frequently targets tasks within the low-precision regime,our study focuses on the high-precision regime crucial for solving *** this work,we study this problem from the following aspects:(i)we analyze the coeffcient matrix that arises in the RFM by studying the distribution of singular values;(ii)we investigate whether the continuous training causes the overfitting issue;(ii)we test direct and iterative methods as well as randomized methods for solving the optimization *** on these results,we find that direct methods are superior to other methods if memory is not an issue,while iterative methods typically have low accuracy and can be improved by preconditioning to some extent.

关键词： Random feature method(RFM) Partial differential equation(PDE) Least-squares problem Direct method Iterative method

来源：评论

学校读者我要写书评

暂无评论

Finding the transcription factor binding locations using novel algorithm segmentation to filtration (S2F)

引用

Journal of Ambient Intelligence and Humanized Computing 2024年第9期15卷 3347-3358页

作者： Theepalakshmi, P. Srinivasulu Reddy, U. Department of Computer Science and Engineering Gandhi Institute of Technology and Management Karnataka Bengaluru India Machine Learning and Data Analytics Lab Center of Excellence in Artificial Intelligence Department of Computer Applications National Institute of Technology Tamilnadu Tiruchirappalli India

The primary aim of identifying the binding motifs in gene regulation is to understand the transcriptional regulation molecular mechanism systematically. In this study, the (, d) motif search issue was considered which entails finding the length motifs which differ by at most d substitutions. However, identifying the high-quality pattern (, d) is challenging. It is intended to address the above problem with motif discovery and handle it using the proposed algorithm S2F (Segmentation to Filtration) based on the qPMS (quorum Planted Motif Search) algorithm model. From the entire DNA sequences, five percent are chosen at random to be used in the motif discovery process. This random sub segment (subseg) portion is split up into base, sub k-mers, and its sizes (motif length ()) are determined by the iterative approach. Corresponding to the sizes of and d (mutations), the k-mers are chosen which participated in filtration techniques and the base k-mer count and frequency are updated. The highest frequency of k-mer is recognized as the motif. The algorithm’s performance was evaluated using the two real datasets Escherichia coli cyclic AMP receptor protein (CRP) and mouse Embryonic Stem Cell (mESC) ChIP-seq (Chromatin Immuno Precipitation) dataset. Results from the experiments show that S2F can identify the motifs and appear faster compared to previous state-of-the-art PMS (Planted Motif Search) and qPMS algorithms. Graphical Abstract: (Figure presented.) © The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024.

关键词： DNA sequences

来源：评论

学校读者我要写书评

暂无评论

Understanding the Expressive Power and Mechanisms of Transformer for Sequence Modeling 38

Understanding the Expressive Power and Mechanisms of Transfo...

引用

38th Conference on Neural Information Processing Systems, NeurIPS 2024

作者： Wang, Mingze Weinan, E. School of Mathematical Sciences Peking University Beijing China Center for Machine Learning Research School of Mathematical Sciences Peking University Beijing China AI for Science Institute Beijing China

We conduct a systematic study of the approximation properties of Transformer for sequence modeling with long, sparse and complicated memory. We investigate the mechanisms through which different components of Transformer, such as the dot-product self-attention, positional encoding and feed-forward layer, affect its expressive power, and we study their combined effects through establishing explicit approximation rates. Our study reveals the roles of critical parameters in the Transformer, such as the number of layers and the number of attention heads. These theoretical insights are validated experimentally and offer natural suggestions for alternative architectures. © 2024 Neural information processing systems foundation. All rights reserved.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Sentiment Analysis on Real-Time Twitter data Using LSTM with Mutually Inclusive Classifiers 8th

Sentiment Analysis on Real-Time Twitter Data Using LSTM with...

引用

8th International Conference on Information and Communication Technology for Competitive Strategies, ICTCS 2023

作者： Soumya, B.J. Swetha, B.N. Meeradevi, Ak. Kanavalli, Anita Shubhangi Singh, Yashashwini Anuritha, L. Singh, Anushka Department of Artificial Intelligence and Data Science Ramaiah Institute of Technology Bengaluru India Department of Artificial Intelligence and Machine Learning Ramaiah Institute of Technology Bengaluru India

ISBN: (纸本)9789819707430

A novel approach to perform sentiment analysis on real-time Twitter data using long short-term memory (LSTM) neural networks with mutually inclusive classifiers. The work leverages the vast amount of publicly available Twitter data, enabling us to analyze the sentiments expressed by users in real-time. The Twitter data by cleaning, tokenizing, and removing stop words to create a high-quality dataset is being preprocessed. The preprocessed data is then fed into the LSTM neural network, which is known for its ability to model long-term dependencies in sequential data. To enhance the sentiment analysis performance, a mutually inclusive classifier framework is proposed. Traditional sentiment analysis models often categorize sentiments into distinct classes, such as positive and negative. In contrast, our mutually inclusive classifiers allow for multiple sentiments to be assigned to a single tweet. This approach is better suited to capture the nuances and complexities of human emotions often expressed on social media platforms like Twitter. The LSTM model with mutually inclusive classifiers using Python and popular deep learning libraries like TensorFlow and Keras is implemented. The model is trained on a labeled dataset, and hyperparameters are tuned to optimize its performance. To evaluate the effectiveness of the approach it uses various metrics, such as accuracy, precision, recall, and F1-score. In the experimental analysis, demonstrated the LSTM-based sentiment analysis model with mutually inclusive classifiers outperforms traditional sentiment analysis models. The findings contribute to the advancement of sentiment analysis techniques and offer valuable applications in understanding and analyzing user sentiments in the dynamic world of social media. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024.

关键词： Sentiment analysis

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：