检索结果-内蒙古大学图书馆

Position: Bayesian Deep Learning is Needed in the Age of Large-Scale AI

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Papamarkou, Theodore Skoularidou, Maria Palla, Konstantina Aitchison, Laurence Arbel, Julyan Dunson, David Filippone, Maurizio Fortuin, Vincent Hennig, Philipp Hernández-Lobato, José Miguel Hubin, Aliaksandr Immer, Alexander Karaletsos, Theofanis Khan, Mohammad Emtiyaz Kristiadi, Agustinus Li, Yingzhen Mandt, Stephan Nemeth, Christopher Osborne, Michael A. Rudner, Tim G.J. Rügamer, David Teh, Yee Whye Welling, Max Wilson, Andrew Gordon Zhang, Ruqi Department of Mathematics The University of Manchester Manchester United Kingdom Eric and Wendy Schmidt Center Broad Institute of MIT and Harvard Cambridge United States Spotify London United Kingdom Computational Neuroscience Unit University of Bristol Bristol United Kingdom Centre Inria de l'Université Grenoble Alpes Grenoble France Department of Statistical Science Duke University United States Statistics Program KAUST Saudi Arabia Helmholtz AI Munich Germany Department of Computer Science Technical University of Munich Munich Germany Munich Center for Machine Learning Munich Germany Tübingen AI Center University of Tübingen Tübingen Germany Department of Engineering University of Cambridge Cambridge United Kingdom Department of Mathematics University of Oslo Oslo Norway Bioinformatics and Applied Statistics Norwegian University of Life Sciences Ås Norway Department of Computer Science ETH Zurich Switzerland Chan Zuckerberg Initiative CA United States Center for Advanced Intelligence Project RIKEN Tokyo Japan Vector Institute Toronto Canada Department of Computing Imperial College London London United Kingdom Department of Computer Science UC Irvine Irvine United States Department of Mathematics and Statistics Lancaster University Lancaster United Kingdom Department of Engineering Science University of Oxford Oxford United Kingdom Center for Data Science New York University New York United States Department of Statistics LMU Munich Munich Germany DeepMind London United Kingdom Department of Statistics University of Oxford Oxford United Kingdom Informatics Institute University of Amsterdam Amsterdam Netherlands Courant Institute of Mathematical Sciences Center for Data Science Computer Science Department New York University New York United States Department of Computer Science Purdue University West Lafayette United States

In the current landscape of deep learning research, there is a predominant emphasis on achieving high predictive accuracy in supervised tasks involving large image and language datasets. However, a broader perspective reveals a multitude of overlooked metrics, tasks, and data types, such as uncertainty, active and continual learning, and scientific data, that demand attention. Bayesian deep learning (BDL) constitutes a promising avenue, offering advantages across these diverse settings. This paper posits that BDL can elevate the capabilities of deep learning. It revisits the strengths of BDL, acknowledges existing challenges, and highlights some exciting research avenues aimed at addressing these obstacles. Looking ahead, the discussion focuses on possible ways to combine large-scale foundation models with BDL to unlock their full potential. © 2024, CC BY-NC-ND.

关键词： Large datasets

Feature selection using stochastic gates 37

学校读者我要写书评

暂无评论

Feature selection using stochastic gates

37th International Conference on Machine Learning, ICML 2020

作者： Yamada, Yutaro Lindenbaum, Ofir Negahban, Sahand Kluger, Yuval Department of Statistics and Data Science Yale University CT United States Program in Applied Mathematics Yale University CT United States School of Medicine Yale University CT United States

ISBN: (纸本)9781713821120

Feature selection problems have been extensively studied in the setting of linear estimation (e.g. LASSO), but less emphasis has been placed on feature selection for non-linear functions. In this study, we propose a method for feature selection in neural network estimation problems. The new procedure is based on probabilistic relaxation of the `0 norm of features, or the count of the number of selected features. Our `0-based regularization relies on a continuous relaxation of the Bernoulli distribution;such relaxation allows our model to learn the parameters of the approximate Bernoulli distributions via gradient descent. The proposed framework simultaneously learns either a nonlinear regression or classification function while selecting a small subset of features. We provide an information-theoretic justification for incorporating Bernoulli distribution into feature selection. Furthermore, we evaluate our method using synthetic and real-life data to demonstrate that our approach outperforms other commonly used methods in both predictive performance and feature selection. Copyright 2020 by the author(s).

关键词： Feature Selection

Empowering Precision Medicine: AI-Driven Schizophrenia Diagnosis via EEG Signals: A Comprehensive Review from 2002-2023

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Jafari, Mahboobeh Sadeghi, Delaram Shoeibi, Afshin Alinejad-Rokny, Hamid Beheshti, Amin García, David López Chen, Zhaolin Acharya, U. Rajendra Gorriz, Juan M. Internship in BioMedical Machine Learning Lab The Graduate School of Biomedical Engineering UNSW Sydney SydneyNSW2052 Australia Data Science and Computational Intelligence Institute University of Granada Spain BioMedical Machine Learning Lab The Graduate School of Biomedical Engineering UNSW Sydney SydneyNSW2052 Australia UNSW Data Science Hub The University of New South Wales SydneyNSW2052 Australia Health Data Analytics Program Centre for Applied Artificial Intelligence Macquarie University Sydney2109 Australia Data Science Lab School of Computing Macquarie University SydneyNSW2109 Australia Monash University Melbourne Australia School of Mathematics Physics and Computing University of Southern Queensland Springfield Australia Department of Psychiatry University of Cambridge United Kingdom

Schizophrenia (SZ) is a prevalent mental disorder characterized by cognitive, emotional, and behavioral changes. Symptoms of SZ include hallucinations, illusions, delusions, lack of motivation, and difficulties in concentration. While the exact causes of SZ remain unproven, factors such as brain injuries, stress, and psychotropic drugs have been implicated in its development. SZ can be classified into different types, including paranoid, disorganized, catatonic, undifferentiated, and residual. Diagnosing SZ involves employing various tools, including clinical interviews, physical examinations, psychological evaluations, the Diagnostic and Statistical Manual of Mental Disorders (DSM), and neuroimaging techniques. Electroencephalography (EEG) recording is a significant functional neuroimaging modality that provides valuable insights into brain function during SZ. However, EEG signal analysis poses challenges for neurologists and scientists due to the presence of artifacts, long-term recordings, and the utilization of multiple channels. To address these challenges, researchers have introduced artificial intelligence (AI) techniques, encompassing conventional machine learning (ML) and deep learning (DL) methods, to aid in SZ diagnosis. This study reviews papers focused on SZ diagnosis utilizing EEG signals and AI methods. The introduction section provides a comprehensive explanation of SZ diagnosis methods and intervention techniques. Subsequently, review papers in this field are discussed, followed by an introduction to the AI methods employed for SZ diagnosis and a summary of relevant papers presented in tabular form. Additionally, this study reports on the most significant challenges encountered in SZ diagnosis, as identified through a review of papers in this field. Future directions to overcome these challenges are also addressed. The discussion section examines the specific details of each paper, culminating in the presentation of conclusions and findings. © 2023,

关键词： Electroencephalography

C2-GaMe: Classification of Cluster Galaxy Membership with Machine Learning

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Farid, Daniel Aung, Han Nagai, Daisuke Farahi, Arya Rozo, Eduardo Program in Applied Mathematics Yale University New HavenCT06511 United States Department of Physics Yale University New HavenCT06520 United States Department of Statistics and Data Science The University of Texas AustinTX78712 United States Department of Physics University of Arizona TucsonAZ85721 United States

We present Classification of Cluster Galaxy Members (C2-GaMe), a classification algorithm based on a suite of machine learning models that differentiates galaxies into orbiting, infalling, and background (interloper) populations, using phase space information as input. We train and test C2-GaMe with the galaxies from UniverseMachine mock catalog based on Multi-Dark Planck 2 N-body simulations. We show that probabilistic classification is superior to deterministic classification in estimating the physical properties of clusters, including density profiles and velocity dispersion. We propose a set of estimators to get an unbiased estimation of cluster properties. We demonstrate that C2-GaMe can recover the distribution of orbiting and infalling galaxies' position and velocity distribution with Copyright © 2022, The Authors. All rights reserved.

关键词： Galaxies

SUPP & MAPP: ADAPTABLE STRUCTURE-BASED REPRESENTATIONS FOR MIR TASKS 21

学校读者我要写书评

暂无评论

SUPP & MAPP: ADAPTABLE STRUCTURE-BASED REPRESENTATIONS FOR M...

21st International Society for Music Information Retrieval Conference, ISMIR 2020

作者： Savard, Claire Bugbee, Erin H. McGuirl, Melissa R. Kinnaird, Katherine M. Department of Physics University of Colorado Boulder United States Department of Biostatistics Brown University United States Division of Applied Mathematics Brown University United States Department of Computer Science and Program in Statistical and Data Sciences Smith College United States

ISBN: (纸本)9780981353708

Accurate and flexible representations of music data are paramount to addressing MIR tasks, yet many of the existing approaches are difficult to interpret or rigid in nature. This work introduces two new song representations for structure-based retrieval methods: Surface Pattern Preservation (SuPP), a continuous song representation, and Matrix Pattern Preservation (MaPP), SuPP’s discrete counterpart. These representations come equipped with several user-defined parameters so that they are adaptable for a range of MIR tasks. Experimental results show MaPP as successful in addressing the cover song task on a set of Mazurka scores, with a mean precision of 0.965 and recall of 0.776. SuPP and MaPP also show promise in other MIR applications, such as novel-segment detection and genre classification, the latter of which demonstrates their suitability as inputs for machine learning problems. © Claire Savard, Erin H. Bugbee, Melissa R. McGuirl, Katherine M. Kinnaird.

关键词： Information retrieval

Bayesian nonparametric hypothesis testing for longitudinal data analysis

学校读者我要写书评

暂无评论

Computational Statistics and data Analysis 2023年 179卷

作者： Pereira, Luz Adriana Gutiérrez, Luis Taylor-Rodríguez, Daniel Mena, Ramsés H. Escuela de Estadística Universidad del Valle Cali Colombia Departamento de Estadística Pontificia Universidad Católica de Chile ANID–Millennium Science Initiative Program–Millennium Nucleus Center for the Discovery of Structures in Complex Data Avenida Vicuña Mackenna 4860 Santiago Chile Department of Mathematics and Statistics Portland State University Portland United States Departamento de Probabilidad y Estadística IIMAS-UNAM Mexico

A Bayesian nonparametric procedure for longitudinal data analysis is proposed. The procedure simultaneously tests for the effects in the mean structure preserving the main effects when interactions are present. The method is highly flexible in that it does not assume a particular distribution for the errors and random effects, as is usually done in longitudinal data analysis. The correlation between the repeated measurements is captured via a Markovian time-dependent Dirichlet process mixture. Specifically, when this latter is represented via a species sampling model with stick-breaking weights, the effect of predictors is driven by the underlying atoms, and the time evolution driven by time-dependent weights. The performance of the proposed method is illustrated using both simulated and real data sets. © 2022 Elsevier B.V.

关键词： Markov processes

Spin-tunable thermoelectric performance in monolayer chromium pnictides

学校读者我要写书评

暂无评论

Physical Review Materials 2022年第6期6卷 064010-064010页

作者： Melania S. Muntini Edi Suprayoga Sasfan A. Wella Iim Fatimah Lila Yuwana Tosawat Seetawan Adam B. Cahaya Ahmad R. T. Nugraha Eddwi H. Hasdeo Department of Physics Faculty of Science and Data Analytics Institut Teknologi Sepuluh Nopember Surabaya 60111 Indonesia Research Center for Quantum Physics National Research and Innovation Agency Tangerang Selatan 15314 Indonesia Center of Excellence on Alternative Energy Research and Development Institution Sakon Nakhon Rajabhat University Sakon Nakhon 47000 Thailand Program of Physics Faculty of Science and Technology Sakon Nakhon Rajabhat University Sakon Nakhon 47000 Thailand Department of Physics Faculty of Mathematics and Natural Sciences Universitas Indonesia Depok 16424 Indonesia Department of Physics and Materials Science University of Luxembourg L-1511 Luxembourg

Historically, finding two-dimensional (2D) magnets is well known to be a difficult task due to the instability against thermal spin fluctuations. Metals are also normally considered poor thermoelectric (TE) materials. Combining intrinsic magnetism in two dimensions with conducting properties, one may expect to get the worst for thermoelectrics. However, we will show this is not always the case. Here, we investigate the spin-dependent TE properties of monolayer chromium pnictides (CrX, where X=P, As, Sb, and Bi) using first-principles calculations of electron- and phonon-energy dispersion, along with the Boltzmann transport formalism under an energy-dependent relaxation time approximation. All the CrX monolayers are dynamically stable and they also exhibit half metallicity with ferromagnetic ordering. Using the spin-valve setup with an antiparallel spin configuration, the half metallicity and ferromagnetism in monolayer CrX enable the manipulation of spin degrees of freedom to tune the TE figure of merit (ZT). At an optimized chemical potential and operating temperature of 500 K, the maximum ZT values (≈0.22, 0.12, and 0.09) with the antiparallel spin-valve setup in CrAs, CrSb, and CrBi improve up to almost twice the original values (ZT≈0.12, 0.08, and 0.05) without the spin-valve configuration. Only in CrP, which is the lightest species and less spin polarized among CrX, the maximum ZT (≈0.34) without the spin-valve configuration is larger than that (≈0.19) with the spin-valve one. We also find that, at 500 K, all the CrX monolayers possess exceptional TE power factors of about 0.02–0.08 W/m K2, which could be one of the best values among 2D conductors.

关键词： First-principles calculations Thermopower Monolayer films Electronic structure

An analysis of reconstruction noise from undersampled 4D flow MRI

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Partin, Lauren Schiavazzi, Daniele E. Long, Carlos A. Sing Department Of Applied And Computational Mathematics And Statistics University Of Notre Dame Notre DameIN United States Institute For Mathematical And Computational Engineering Institute For Biological And Medical Engineering Pontificia Universidad Católica De Chile Santiago Chile ANID - Millennium Science Initiative Program Millennium Nucleus Center For The Discovery Of Structures In Complex Data Chile ANID - Millennium Science Initiative Program Millennium Nucleus Center For Cardiovascular Magnetic Resonance Chile

Novel Magnetic Resonance (MR) imaging modalities can quantify hemodynamics but require long acquisition times, precluding its widespread use for early diagnosis of cardiovascular disease. To reduce the acquisition times, reconstruction methods from undersampled measurements are routinely used, that leverage representations designed to increase image compressibility. Reconstructed anatomical and hemodynamic images may present visual artifacts. Although some of these artifact are essentially reconstruction errors, and thus a consequence of undersampling, others may be due to measurement noise or the random choice of the sampled frequencies. Said otherwise, a reconstructed image becomes a random variable, and both its bias and its covariance can lead to visual artifacts;the latter leads to spatial correlations that may be misconstrued for visual information. Although the nature of the former has been studied in the literature, the latter has not received as much attention. In this study, we investigate the theoretical properties of the random perturbations arising from the reconstruction process, and perform a number of numerical experiments on simulated and MR aortic flow. Our results show that the correlation length remains limited to two to three pixels when a Gaussian undersampling pattern is combined with recovery algorithms based on 1-norm minimization. However, the correlation length may increase significantly for other undersampling patterns, higher undersampling factors (i.e., 8× or 16× compression), and different reconstruction methods. © 2022, CC BY-NC-ND.

关键词： Magnetic resonance

data science in science: Special Issue on data science in the Brain sciences

学校读者我要写书评

暂无评论

data science in science 2023年第1期2卷

作者： Carolina Euan Mark B. Fiecas Hernando Ombao David S. Matteson a Department of Mathematics and Statistics Lancaster University b Division of Biostatistics University of Minnesota c Statistics Program King Abdullah University of Science and Technology d Department of Statistics and Data Science Cornell University