检索结果-内蒙古大学图书馆

Sampling from Mean-Field Gibbs Measures via Diffusion Processes

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Alaoui, Ahmed El Montanari, Andrea Sellke, Mark Department of Statistics and Data Science Cornell University United States Department of Statistics Department of Mathematics Stanford University United States Department of Statistics Harvard University United States

We consider Ising mixed p-spin glasses at high-temperature and without external field, and study the problem of sampling from the Gibbs distribution µ in polynomial time. We develop a new sampling algorithm with complexity of the same order as evaluating the gradient of the Hamiltonian and, in particular, at most linear in the input size. We prove that, at sufficiently high-temperature, it produces samples from a distribution µalg which is close in normalized Wasserstein distance to µ. Namely, there exists a coupling of µ and µalg such that if (x, xalg) ∈ {−1, +1}n × {−1, +1}n is a pair drawn from this coupling, then n−1 E{kx − xalgk22} = on(1). For the case of the Sherrington-Kirkpatrick model, our algorithm succeeds in the full replica-symmetric phase. Previously, [ABXY22, AJK+23] showed that Glauber dynamics succeeds in sampling under a stronger assumption on the temperature. (However, these works prove sampling in total variation distance.) We complement this result with a negative one for sampling algorithms satisfying a certain ‘stability’ property, which is verified by many standard techniques. No stable algorithm can approximately sample at temperatures below the onset of shattering, even under the normalized Wasserstein metric. Further, no algorithm can sample at temperatures below the onset of replica symmetry breaking. Our sampling method implements a discretized version of a diffusion process that has become recently popular in machine learning under the name of ‘denoising diffusion.’ We derive the same process from the general construction of stochastic localization. Implementing the diffusion process requires to efficiently approximate the mean of the tilted measure. To this end, we use an approximate message passing algorithm that, as we prove, achieves sufficiently accurate mean estimation. Copyright © 2023, The Authors. All rights reserved.

关键词： Learning algorithms

STABILITY AND STATISTICAL INFERENCE FOR SEMIDISCRETE OPTIMAL TRANSPORT MAPS

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Ritwik, Sadhu Ziv, Goldfeld Kengo, Kato Department of Statistics and Data Science Cornell University United States School of Electrical and Computer Engineering Cornell University United States

We study statistical inference for the optimal transport (OT) map (also known as the Brenier map) from a known absolutely continuous reference distribution onto an unknown finitely discrete target distribution. We derive limit distributions for the Lp-error with arbitrary p ∈ [1,∞) and for linear functionals of the empirical OT map, together with their moment convergence. The former has a non-Gaussian limit, whose explicit density is derived, while the latter attains asymptotic normality. For both cases, we also establish consistency of the nonparametric bootstrap. The derivation of our limit theorems relies on new stability estimates of functionals of the OT map with respect to the dual potential vector, which may be of independent interest. We also discuss applications of our limit theorems to the construction of confidence sets for the OT map and inference for a maximum tail correlation. Finally, we show that, while the empirical OT map does not possess nontrivial weak limits in the L2 space, it satisfies a central limit theorem in a dual Hölder space, and the Gaussian limit law attains the asymptotic efficiency bound. Copyright © 2023, The Authors. All rights reserved.

关键词：

An Exploration of Interpolation - Machine Learning Model for Climate Model Downscaling Under the Limitation of data Quantity

学校读者我要写书评

暂无评论

An Exploration of Interpolation - Machine Learning Model for...

2023 International Technical Conference on Circuits/Systems, computers, and Communications, ITC-CSCC 2023

作者： Prathom, Chotirose Champrasert, Paskorn Chiang Mai University Data Science Consortium Faculty of Engineering Chiang Mai Thailand Chiang Mai University Faculty of Engineering Department of Computer Engineering Chiang Mai Thailand

ISBN: (纸本)9798350326413

In order to response to climate change, climate model is essential for future prediction, which is advantage for adaptation planning. With the fact that the output of the climate model is in a coarse spatial resolution, it is poorly representing data on local level and requires a downscaling process to increase its resolution. To perform the downscaling, high amount of historical observed data is necessary for calibrating the model output. However, the data may not be available in some area, also Thailand. To overcome the limitation of data Quantity, four combination methods between two interpolation (Inverse Distance Weight and Triangular Irregular Network) and two machine learning models (Artificial Neural Network and Gradient Boosting Regression Tree) are experimented to explore the proper one for downscaling data on Thailand area. By considering the performance of both interpolation and machine learning model, the combination between Inverse Distance Weight and Artificial Neural Network shows the best performance in down scaling process under data Quantity limitation. © 2023 IEEE.

关键词： Climate models

Multi-Modal Imaging Genomics Transformer: Attentive Integration of Imaging with Genomic Biomarkers for Schizophrenia Classification

学校读者我要写书评

暂无评论

Multi-Modal Imaging Genomics Transformer: Attentive Integrat...

IEEE International Symposium on Biomedical Imaging

作者： Nagur Shareef Shaik Teja Krishna Cherukuri Vince Calhoun Dong Hye Ye Department of Computer Science Georgia State University Translational Research in Neuroimaging and Data Science (TReNDS) Georgia State Georgia Tech Emory

ISBN: (数字)9798331520526

ISBN: (纸本)9798331520533

Schizophrenia (SZ) is a severe brain disorder marked by diverse cognitive impairments, abnormalities in brain structure, function, and genetic factors. Its complex symptoms and overlap with other psychiatric conditions complicate diagnosis, necessitating advanced systems. Existing research studies have mostly focused on imaging data, such as structural and functional MRI, for SZ diagnosis. There has been less focus on the integration of genomic features despite their potential in identifying heritable SZ traits. This study introduces a Multi-modal Imaging Genomics Transformer (MIGTrans) that integrates genomics with structural and functional imaging to identify SZ-related abnormalities. MIGTrans improved SZ classification accuracy to 86.05%, offering clear interpretations and identifying key genomic and brain morphological/connectivity patterns linked to SZ.

关键词： Brain Accuracy Magnetic resonance imaging Decision making Genomics Imaging Schizophrenia Transformers Bioinformatics Biomedical imaging

Boli: A dataset for understanding stuttering experience and analyzing stuttered speech

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Batra, Ashita Narang, Mannas Sharma, Neeraj Kumar Das, Pradip K. Department of Computer Science and Engineering Mehta Family School of Data Science and AI Indian Institute of Technology Guwahati Guwahati781039 India

There is a growing need for diverse, high-quality stuttered speech data, particularly in the context of Indian languages. This paper introduces Project Boli, a multi-lingual stuttered speech dataset designed to advance scientific understanding and technology development for individuals who stutter, particularly in India. The dataset constitutes (a) anonymized metadata (gender, age, country, mother tongue) and responses to a questionnaire about how stuttering affects their daily lives, (b) captures both read speech (using the Rainbow Passage) and spontaneous speech (through image description tasks) for each participant and (c) includes detailed annotations of five stutter types: blocks, prolongations, interjections, sound repetitions and word repetitions. We present a comprehensive analysis of the dataset, including the data collection procedure, experience summarization of people who stutter, severity assessment of stuttering events and technical validation of the collected data. The dataset is released as an open access to further speech technology development. © 2025, CC BY.

关键词： data assimilation

StudentCoin Price Prediction and Relation to Blockchain Education

学校读者我要写书评

暂无评论

StudentCoin Price Prediction and Relation to Blockchain Educ...

International Conference on Computing for Sustainable Global Development (INDIACom)

作者： Rashmeet Kaur Sonia Carmela Iorio Department of Economics Science & Statistics University of Naples Federico II Naples Italy Yogananda School of AI Computer and Data Science Shoolini University Solan India

The newly introduced cryptocurrency called StudentCoin which is identified by the acronym STC is a blockchain-based project. It is the first platform in the world for easy tokenization. It allows the users to learn about cryptocurrencies, blockchain technology and their use cases. Users can create their own tokens, coins and also trade them on the same terminal. This paper uses easy-to-understand modeling and simulation techniques to make price predictions for blockchain education-oriented cryptocurrencies like StudentCoin. We have used the Monte Carlo simulation technique to determine if the price of StudentCoin is expected to increase within a month, and that combining education with blockchain and cryptocurrency platforms is beneficial. The study also aims to find out if college students are aware of cryptocurrencies and see if it's possible to combine education and cryptocurrency. The paper contributes to promoting awareness among university students about responsible investment practices in the cryptocurrency market.

关键词： Monte Carlo methods Terminology Education Predictive models Tokenization Cryptocurrency Blockchains

NFL Ghosts: A framework for evaluating defender positioning with conditional density estimation

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Yurko, Ronald Nguyen, Quang Pelechrinis, Konstantinos Department of Statistics & Data Science Carnegie Mellon University United States Department of Informatics and Networked Systems University of Pittsburgh United States

Player attribution in American football remains an open problem due to the complex nature of twenty-two players interacting on the field, but the granularity of player tracking data provides ample opportunity for novel approaches. In this work, we introduce the first public framework to evaluate spatial and trajectory tracking data of players relative to a baseline distribution of "ghost" defenders. We demonstrate our framework in the context of modeling the nearest defender positioning at the moment of catch. In particular, we provide estimates of how much better or worse their observed positioning and trajectory compared to the expected play value of ghost defenders. Our framework leverages high-dimensional tracking data features through flexible random forests for conditional density estimation in two ways: (1) to model the distribution of receiver yards gained enabling the estimation of within-play expected value, and (2) to model the 2D spatial distribution of baseline ghost defenders. We present novel metrics for measuring player and team performance based on tracking data, and discuss challenges that remain in extending our framework to other aspects of American football. © 2024, CC BY.

关键词： Clustering algorithms

Recent Advancements in Hyperspectral Face Recognition: A Review 13

学校读者我要写书评

暂无评论

Recent Advancements in Hyperspectral Face Recognition: A Rev...

13th International Conference on Computing Communication and Networking Technologies, ICCCNT 2022

作者： Das, Samiran Chatterjee, Sujoy JIS University Centre for Data Science JIS Institute for Advanced Studies and Research Kolkata India Amity University Department of Computer Science and Engineering Kolkata India

ISBN: (数字)9781665452625

ISBN: (纸本)9781665452625

The increasing demand for high-accuracy face recognition methods and the rapid advancement in instrumentation has made hyperspectral face recognition a viable alternative to traditional face recognition. The high-security biometric application scenarios, which demand high accuracy of recognition are switching to hyperspectral imaging due to its robustness against spoofing, illumination, pose variation, etc. The hundreds of visible and near infrared (NIR) images contained in a hyperspectral face cube provide sufficient information content, leading to more accurate recognition. However, the gigantic volume of information and the computational challenges call for efficient learning strategies for the recognition task. Researchers have introduced several feature extraction, supervised learning, deep learning, and pattern recognition approaches to carry out face recognition from hyperspectral face cubes. This work summarized the state-of-the-art approaches for hyperspectral face recognition in detail and highlight their pros and cons. The work also discusses future directions and challenges in this domain. © 2022 IEEE.

关键词： Biometrics

Using Long Short-Term Memory Units for Time Series Forecasting 2

学校读者我要写书评

暂无评论

Using Long Short-Term Memory Units for Time Series Forecasti...

2nd IEEE International Conference on Futuristic Technologies, INCOFT 2023

作者： Yadav, Dhyan Sahoo, Laxman Mandal, Sanjeev Kumar Ravivarman, G. Vijayaraghavan, P. Prasad, B. Maharishi School of Engineering and Technology Maharishi University of Information Technology Uttar Pradesh India Vivekananda Global University Department of Computer Science & Engineering Jaipur India Department of Computer Science and Information Technology Karnataka Bangalore India Department of Electrical and Electronics Engineering Karpagam Academy of Higher Education Coimbatore641021 India Prince Shri Venkateshwara Padmavathy Engineering College Department of Mechanical Engineering Chennai127 India Vishwakarma Institute of Information Technology Department of Artificial Intelligence & Data Science Pune India

ISBN: (纸本)9798350308846

This paper specializes in using long short-time period memory (LSTM) networks for time collection forecasting. LSTM is a form of artificial recurrent neural network (RNN) that is especially properly-applicable to analyze from enjoy over long sequences of statistics. Not like conventional recurrent networks, LSTM networks can technique input sequences in each instruction, while also allowing for the protection of lengthy-time period dependencies and rapid convergence. It's far shown that these characteristics of LSTM networks allow them to be used efficaciously for forecasting time series consisting of inventory performance and electricity tendencies. Strategies inclusive of feeding the LSTM network data in a time step layout, the use of peephole connections, stacking a couple of layers and early stopping are discussed. Results are then presented and mentioned, demonstrating that LSTM networks can research complex lengthy-time period dependencies in time series facts and correctly forecast destiny values. © 2023 IEEE.

关键词： Complex Dependencies Destiny Research Values