检索结果-内蒙古大学图书馆

Monitoring AI-Modified Content at Scale: A Case Study on the Impact of ChatGPT on AI Conference Peer Reviews

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Liang, Weixin Izzo, Zachary Zhang, Yaohui Lepp, Haley Cao, Hancheng Zhao, Xuandong Chen, Lingjiao Ye, Haotian Liu, Sheng Huang, Zhi McFarland, Daniel A. Zou, James Y. Department of Computer Science Stanford University United States Machine Learning Department NEC Labs America United States Department of Electrical Engineering Stanford University United States Graduate School of Education Stanford University United States Department of Management Science and Engineering Stanford University United States Department of Computer Science UC Santa Barbara United States Department of Biomedical Data Science Stanford University United States Department of Sociology Stanford University United States Graduate School of Business Stanford University United States

We present an approach for estimating the fraction of text in a large corpus which is likely to be substantially modified or produced by a large language model (LLM). Our maximum likelihood model leverages expert-written and AI-generated reference texts to accurately and efficiently examine real-world LLM-use at the corpus level. We apply this approach to a case study of scientific peer review in AI conferences that took place after the release of ChatGPT: ICLR 2024, NeurIPS 2023, CoRL 2023 and EMNLP 2023. Our results suggest that between 6.5% and 16.9% of text submitted as peer reviews to these conferences could have been substantially modified by LLMs, i.e. beyond spell-checking or minor writing updates. The circumstances in which generated text occurs offer insight into user behavior: the estimated fraction of LLM-generated text is higher in reviews which report lower confidence, were submitted close to the deadline, and from reviewers who are less likely to respond to author rebuttals. We also observe corpus-level trends in generated text which may be too subtle to detect at the individual level, and discuss the implications of such trends on peer review. We call for future interdisciplinary work to examine how LLM use is changing our information and knowledge practices. Copyright © 2024, The Authors. All rights reserved.

关键词： Maximum likelihood

The three stages of learning dynamics in high-dimensional kernel methods

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Ghosh, Nikhil Mei, Song Yu, Bin Department of Statistics UC Berkeley Department of Electrical Engineering and Computer Science UC Berkeley Center for Computational Biology UC Berkeley Weill Neurohub Investigator

To understand how deep learning works, it is crucial to understand the training dynamics of neural networks. Several interesting hypotheses about these dynamics have been made based on empirically observed phenomena, but there exists a limited theoretical understanding of when and why such phenomena occur. In this paper, we consider the training dynamics of gradient flow on kernel least-squares objectives, which is a limiting dynamics of SGD trained neural networks. Using precise high-dimensional asymptotics, we characterize the dynamics of the fitted model in two "worlds": in the Oracle World the model is trained on the population distribution and in the Empirical World the model is trained on a sampled dataset. We show that under mild conditions on the kernel and L2 target regression function the training dynamics undergo three stages characterized by the behaviors of the models in the two worlds. Our theoretical results also mathematically formalize some interesting deep learning phenomena. Specifically, in our setting we show that SGD progressively learns more complex functions and that there is a "deep bootstrap" phenomenon: during the second stage, the test error of both worlds remain close despite the empirical training error being much smaller. Finally, we give a concrete example comparing the dynamics of two different kernels which shows that faster training is not necessary for better generalization. © 2021, CC BY.

关键词： Dynamics

Do you see what I see? A Cross-cultural Comparison of Social Impressions of Faces 42

学校读者我要写书评

暂无评论

Do you see what I see? A Cross-cultural Comparison of Social...

42nd Annual Meeting of the Cognitive science Society: Developing a Mind: Learning in Humans, Animals, and Machines, CogSci 2020

作者： Song, Amanda Hu, Weifeng Yavav, Devendra Pratap Wen, Fangfang Zuo, Bin Vul, Edward Cottrell, Garrison Cognitive Science UC San Diego United States Computer Science and Engineering UC San Diego United States Psychology Central China Normal University China Psychology UC San Diego United States

Research has suggested that social impressions of faces made by Western and Eastern people have different underlying dimensionalities. However, the individual level consistency, the group-level agreement of rater groups, and the interactions between face ethnicity, rater ethnicity, and social impression traits remain largely unknown. In this paper, we perform a large-scale data-driven cross-cultural study of facial impressions, and illustrate the idiosyncrasies and similarities behind Caucasian and Asian participants in their social impressions of faces from both ethnicity groups. Our study illustrates multiple interesting findings: (1) Asians rate faces lower on most positive traits, compared with Caucasian raters, and they have more diverse opinions than Caucasians. (2) Caucasian faces receive higher average ratings on social impression traits related to warmth due to the preponderance of smiles in Caucasian images, but similar mean scores on traits related to capability, compared to Asian faces. (3) Caucasians and Asians disagree most on capability related traits, especially on "responsible" and "successful." Opinions on these two traits diverge more on Asian than on Caucasian faces. Our findings provide new insights on the nuances of cross-cultural differences in social impressions of faces. © 2020 The Author(s)

关键词： cross-cultural comparison face perception First impressions large scale online experiment statistical analysis

Deep and shallow data science for multi-scale optical neuroscience

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Mishne, Gal Charles, Adam Halıcıoğlu Data Science Institute Department of Electrical and Computer Engineering the Neurosciences Graduate Program UC San Diego 9500 Gilman Drive La Jolla CA92093 United States Department of Biomedical Engineering Kavli Neuroscience Discovery Institute Center for Imaging Science Department of Neuroscience Mathematical Institute for Data Science Johns Hopkins University BaltimoreMD21287 United States

Optical imaging of the brain has expanded dramatically in the past two decades. New optics, indicators, and experimental paradigms are now enabling in-vivo imaging from the synaptic to the cortex-wide scales. To match the resulting flood of data across scales, computational methods are continuously being developed to meet the need of extracting biologically relevant information. In this pursuit, challenges arise in some domains (e.g., SNR and resolution limits in micron-scale data) that require specialized algorithms. These algorithms can, for example, make use of state-of-the-art machine learning to maximally learn the details of a given scale to optimize the processing pipeline. In contrast, other methods, however, such as graph signal processing, seek to abstract away from some of the details that are scale-specific to provide solutions to specific sub-problems common across scales of neuroimaging. Here we discuss limitations and tradeoffs in algorithmic design with the goal of identifying how data quality and variability can hamper algorithm use and dissemination. © 2024, CC BY.

关键词： Neuroimaging

A Mixing Time Lower Bound for a Simplified Version of BART

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Ronen, Omer Saarinen, Theo Tan, Yan Shuo Duncan, James Yu, Bin Department of Statistics UC Berkeley United States Department of Electrical Engineering and Computer Sciences UC Berkeley United States Center for Computational Biology UC Berkeley United States Department of Statistics and Data Science National University of Singapore Singapore Microsoft Research United States Group in Biostatistics UC Berkeley United States

Bayesian Additive Regression Trees (BART) is a popular Bayesian non-parametric regression algorithm. The posterior is a distribution over sums of decision trees, and predictions are made by averaging approximate samples from the posterior. The combination of strong predictive performance and the ability to provide uncertainty measures has led BART to be commonly used in the social sciences, biostatistics, and causal inference. BART uses Markov Chain Monte Carlo (MCMC) to obtain approximate posterior samples over a parameterized space of sums of trees, but it has often been observed that the chains are slow to mix. In this paper, we provide the first lower bound on the mixing time for a simplified version of BART in which we reduce the sum to a single tree and use a subset of the possible moves for the MCMC proposal distribution. Our lower bound for the mixing time grows exponentially with the number of data points. Inspired by this new connection between the mixing time and the number of data points, we perform rigorous simulations on BART. We show qualitatively that BART’s mixing time increases with the number of data points. The slow mixing time of the simplified BART suggests a large variation between different runs of the simplified BART algorithm and a similar large variation is known for BART in the literature. This large variation could result in a lack of stability in the models, predictions and posterior intervals obtained from the BART MCMC samples. Our lower bound and simulations suggest increasing the number of chains with the number of data points. Copyright © 2022, The Authors. All rights reserved.

关键词： Decision trees

Cluster-and-Conquer: A Framework for Time-Series Forecasting

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Pathak, Reese Sen, Rajat Rao, Nikhil Erichson, N. Benjamin Jordan, Michael I. Dhillon, Inderjit S. Department of Electrical Engineering and Computer Sciences UC Berkeley Google Research Amazon School of Engineering University of Pittsburgh Department of Statistics UC Berkeley Department of Computer Science UT Austin

We propose a three-stage framework for forecasting high-dimensional time-series data. Our method first estimates parameters for each univariate time series. Next, we use these parameters to cluster the time series. These clusters can be viewed as multivariate time series, for which we then compute parameters. The forecasted values of a single time series can depend on the history of other time series in the same cluster, accounting for intra-cluster similarity while minimizing potential noise in predictions by ignoring inter-cluster effects. Our framework—which we refer to as "cluster-and-conquer"—is highly general, allowing for any time-series forecasting and clustering method to be used in each step. It is computationally efficient and embarrassingly *** motivate our framework with a theoretical analysis in an idealized mixed linear regression setting, where we provide guarantees on the quality of the estimates. We accompany these guarantees with experimental results that demonstrate the advantages of our framework: when instantiated with simple linear autoregressive models, we are able to achieve state-of-the-art results on several benchmark datasets, sometimes outperforming deep-learning-based approaches. © 2021, CC BY.

关键词： Time series

When is the estimated propensity score better? High-dimensional analysis and bias correction

学校读者我要写书评

暂无评论

arXiv 2023年

作者： Su, Fangzhou Mou, Wenlong Ding, Peng Wainwright, Martin J. Department of Electrical Engineering and Computer Sciences United States Department of Statistics UC Berkeley United States Department of Mathematics Lab for Information and Decision Systems and Statistics and Data Science Center Massachusetts Institute of Technology United States

Anecdotally, using an estimated propensity score is superior to the true propensity score in estimating the average treatment effect based on observational data. However, this claim comes with several qualifications: it holds only if propensity score model is correctly specified and the number of covariates d is small relative to the sample size n. We revisit this phenomenon by studying the inverse propensity score weighting (IPW) estimator based on a logistic model with a diverging number of covariates. We first show that the IPW estimator based on the estimated propensity score is consistent and asymptotically normal with smaller variance than the oracle IPW estimator (using the true propensity score) if and only if n & d2. We then propose a debiased IPW estimator that achieves the same guarantees in the regime n & d3/2. Our proofs rely on a novel non-asymptotic decomposition of the IPW error along with careful control of the higher order terms. Copyright © 2023, The Authors. All rights reserved.

关键词：

Position: Bayesian Deep Learning is Needed in the Age of Large-Scale AI 41

学校读者我要写书评

暂无评论

Position: Bayesian Deep Learning is Needed in the Age of Lar...

41st International Conference on Machine Learning, ICML 2024

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. Copyright 2024 by the author(s)

关键词： Contrastive Learning

A Novel Aerial-Aquatic Locomotion Robot with Variable Stiffness Propulsion Module

学校读者我要写书评

暂无评论

arXiv 2024年

作者： Hu, Junzhe Chen, Pengyu Feng, Tianxiang Wen, Yuxuan Wu, Ke Dong, Janet The UC Center for Robotics Research College of Engineering and Applied Science University of Cincinnati CincinnatiOH United States The Robotics Institute School of Computer Science Carnegie Mellon University PittsburghPA United States Robotics department Mohamed bin Zayed University of Artificial Intelligence Masdar City Abu Dhabi United Arab Emirates

In recent years, the development of robots capable of operating in both aerial and aquatic environments has gained significant attention. This study presents the design and fabrication of a novel aerial-aquatic locomotion robot (AALR). Inspired by the diving beetle, the AALR incorporates a biomimetic propulsion mechanism with power and recovery strokes. The variable stiffness propulsion module (VSPM) uses low melting point alloy (LMPA) and variable stiffness joints (VSJ) to achieve efficient aquatic locomotion while reduce harm to marine life. The AALR's innovative design integrates the VSPM into the arms of a traditional quadrotor, allowing for effective aerial-aquatic locomotion. The VSPM adjusts joint stiffness through temperature control, meeting locomotion requirements in both aerial and aquatic modes. A dynamic model for the VSPM was developed, with optimized dimensional parameters to increase propulsion force. Experiments focused on aquatic mode analysis and demonstrated the AALR's swimming capability, achieving a maximum swimming speed of 77 mm/s underwater. The results confirm the AALR's effective performance in water environment, highlighting its potential for versatile, eco-friendly operations. Copyright © 2024, The Authors. All rights reserved.

关键词： Aircraft propulsion