检索结果-内蒙古大学图书馆

A Probabilistically Motivated learning Rate Adaptation for Stochastic Optimization

学校读者我要写书评

暂无评论

arXiv 2021年

作者： de Roos, Filip Jidling, Carl Wills, Adrian Schön, Thomas B. Hennig, Philipp Department of Computer Science University of Tübingen Germany Max Planck Institute for Intelligent Systems Tübingen Germany Department of Information Technology Uppsala University Sweden School of Engineering University of Newcastle Australia

Machine learning practitioners invest significant manual and computational resources in finding suitable learning rates for optimization algorithms. We provide a probabilistic motivation, in terms of Gaussian inference, for popular stochastic first-order methods. As an important special case, it recovers the Polyak step with a general metric. The inference allows us to relate the learning rate to a dimensionless quantity that can be automatically adapted during training by a control algorithm. The resulting meta-algorithm is shown to adapt learning rates in a robust manner across a large range of initial values when applied to deep learning benchmark problems. © 2021, CC BY.

关键词： learning algorithms

Beta Shapley: a Unified and Noise-reduced Data Valuation Framework for Machine learning

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Kwon, Yongchan Zou, James Department of Biomedical Data Science Stanford University United States

Data Shapley has recently been proposed as a principled framework to quantify the contribution of individual datum in machine learning. It can effectively identify helpful or harmful data points for a learning algorithm. In this paper, we propose Beta Shapley, which is a substantial generalization of Data Shapley. Beta Shapley arises naturally by relaxing the efficiency axiom of the Shapley value, which is not critical for machine learning settings. Beta Shapley unifies several popular data valuation methods and includes data Shapley as a special case. Moreover, we prove that Beta Shapley has several desirable statistical properties and propose efficient algorithms to estimate it. We demonstrate that Beta Shapley outperforms state-of-the-art data valuation methods on several downstream ML tasks such as: 1) detecting mislabeled training data;2) learning with subsamples;and 3) identifying points whose addition or removal have the largest positive or negative impact on the model. © 2021, CC BY-SA.

关键词： learning algorithms

A brief survey of associations between meta-learning and general AI

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Peng, Huimin

This paper briefly reviews the history of meta-learning and describes its contribution to general AI. Meta-learning improves model generalization capacity and devises general algorithms applicable to both in-distribution and out-of-distribution tasks potentially. General AI replaces task-specific models with general algorithmic systems introducing higher level of automation in solving diverse tasks using AI. We summarize main contributions of meta-learning to the developments in general AI, including memory module, meta-learner, coevolution, curiosity, forgetting and AI-generating algorithm. We present connections between meta-learning and general AI and discuss how meta-learning can be used to formulate general AI algorithms. © 2021, CC BY.

关键词： learning algorithms

Reproducibility Issues for BERT-based Evaluation Metrics

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Chen, Yanran Belouadi, Jonas Eger, Steffen Computer Science Department Technical University of Darmstadt Germany NLLG Faculty of Technology Bielefeld University Germany

Reproducibility is of utmost concern in machine learning and natural language processing (NLP). In the field of natural language generation (especially machine translation), the seminal paper of Post (2018) has pointed out problems of reproducibility of the dominant metric, BLEU, at the time of publication. Nowadays, BERT-based evaluation metrics considerably outperform BLEU. In this paper, we ask whether results and claims from four recent BERT-based metrics can be reproduced. We find that reproduction of claims and results often fails because of (i) heavy undocumented preprocessing involved in the metrics, (ii) missing code and (iii) reporting weaker results for the baseline metrics. (iv) In one case, the problem stems from correlating not to human scores but to a wrong column in the csv file, inflating scores by 5 points. Motivated by the impact of preprocessing, we then conduct a second study where we examine its effects more closely (for one of the metrics). We find that preprocessing can have large effects, especially for highly inflectional languages. In this case, the effect of preprocessing may be larger than the effect of the aggregation mechanism (e.g., greedy alignment vs. Word Mover Distance). © 2022, CC BY.

关键词： learning algorithms

Optimal quantum reservoir computing for the NISQ era

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Colomer, Laia Domingo Carlo, G. Borondo, F. Campus de Cantoblanco Nicolás Cabrera 13-15 Madrid28049 Spain Departamento de Química Universidad Autónoma de Madrid Cantoblanco Madrid28049 Spain Grupo de Sistemas Complejos Universidad Politécnica de Madrid Madrid28035 Spain Comisión Nacional de Energía Atómica CONICET Departamento de Física Av. del Libertador 8250 Buenos Aires1429 Argentina

Universal fault-tolerant quantum computers require millions of qubits with low error rates. Since this technology is years ahead, noisy intermediate-scale quantum (NISQ) computation is receiving tremendous interest. In this setup, quantum reservoir computing is a relevant machine learning algorithm. Its simplicity of training and implementation allows to perform challenging computations on today available machines. In this Letter, we provide a criterion to select optimal quantum reservoirs, requiring few and simple gates. Our findings demonstrate that they render better results than other commonly used models with significantly less gates, and also provide insight on the theoretical gap between quantum reservoir computing and the theory of quantum states complexity. © 2022, CC BY-NC-SA.

关键词： learning algorithms

Low-Regret Active learning

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Baykal, Cenk Liebenwein, Lucas Gal, Oren Feldman, Dan Rus, Daniela MIT United States University of Haifa

We develop an online learning algorithm for identifying unlabeled data points that are most informative for training (i.e., active learning). By formulating the active learning problem as the prediction with sleeping experts problem, we provide a regret minimization framework for identifying relevant data with respect to any given definition of informativeness. Motivated by the successes of ensembles in active learning, we define regret with respect to an omnipotent algorithm that has access to an infinity large ensemble. At the core of our work is an efficient algorithm for sleeping experts that is tailored to achieve low regret on easy instances while remaining resilient to adversarial ones. Low regret implies that we can be provably competitive with an ensemble method without the computational burden of having to train an ensemble. This stands in contrast to state-of-the-art active learning methods that are overwhelmingly based on greedy selection, and hence cannot ensure good performance across problem instances with high amounts of noise. We present empirical results demonstrating that our method (i) instantiated with an informativeness measure consistently outperforms its greedy counterpart and (ii) reliably outperforms uniform sampling on real-world scenarios. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

Non-stationary reinforcement learning without prior knowledge: An optimal black-box approach

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wei, Chen-Yu Luo, Haipeng University of Southern California

We propose a black-box reduction that turns a certain reinforcement learning algorithm with optimal regret in a (near-)stationary environment into another algorithm with optimal dynamic regret in a non-stationary environment, importantly without any prior knowledge on the degree of non-stationarity. By plugging different algorithms into our black-box, we provide a list of examples showing that our approach not only recovers recent results for (contextual) multi-armed bandits achieved by very specialized algorithms, but also significantly improves the state of the art for (generalized) linear bandits, episodic MDPs, and infinite-horizon MDPs in various ways. Specifically, in most cases our algorithm achieves the optimal dynamic regret Oe(min{√LT, ∆1/3T2/3}) where T is the number of rounds and L and ∆ are the number and amount of changes of the world respectively, while previous works only obtain suboptimal bounds and/or require the knowledge of L and ∆. © 2021, CC0.

关键词： learning algorithms

Understand and improve contrastive learning methods for visual representation: A review

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Liu, Ran Machine Learning program - ECE Georgia Tech

Traditional supervised learning methods are hitting a bottleneck [33] because of their dependency on expensive manually labeled data and their weaknesses such as limited generalization ability and vulnerability to adversarial attacks [17, 51, 33]. A promising alternative, self-supervised learning, as a type of unsupervised learning, has gained popularity because of its potential to learn effective data representations without manual labeling. Among self-supervised learning algorithms, contrastive learning has achieved state-of-the-art performance in several fields of research, including but not limited to computer vision [7, 20, 39, 53], natural language processing [29, 30], and biomedical image processing [6, 3, 12, 1]. This literature review aims to provide an up-to-date analysis of researchers’ efforts to understand the key components and the limitations of self-supervised learning. Section 2 explains a few basic concepts of self-supervied learning. Section 3 discusses how pretext tasks are used in self-supervised learning, and how being invariant to image transformations is better than being covariant to image transformations [35]. Then, in section 4.1, we describe a recent contrastive learning method SimCLR [7] and in section 4.2 we introduce attempts to improve contrastive methods by altering its key components [26]. Section 5.1 presents the model bias [44] in contrastive learning, and is followed by section 5.2 which discusses possible methods to reduce model bias [46, 49] in current self-supervised learning methods. © 2021, CC BY.

关键词： learning algorithms

learning the Hypotheses Space from Data through a U-Curve Algorithm

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Marcondes, Diego Simonis, Adilson Barrera, Junior Institute of Mathematics and Statistics University of São Paulo São Paulo SP05508-090 Brazil

This paper proposes a data-driven systematic, consistent and non-exhaustive approach to Model Selection, that is an extension of the classical agnostic PAC learning model. In this approach, learning problems are modeled not only by a hypothesis space H, but also by a learning Space L(H), a poset of subspaces of H, which covers H and satisfies a property regarding the VC dimension of related subspaces, that is a suitable algebraic search space for Model Selection algorithms. Our main contributions are a data-driven general learning algorithm to perform implicitly regularized Model Selection on L(H) and a framework under which one can, theoretically, better estimate a target hypothesis with a given sample size by properly modeling L(H) and employing high computational power. A remarkable consequence of this approach are conditions under which a non-exhaustive search of L(H) can return an optimal solution. The results of this paper lead to a practical property of Machine learning, that the lack of experimental data may be mitigated by a high computational capacity. In a context of continuous popularization of computational power, this property may help understand why Machine learning has become so important, even where data is expensive and hard to get. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms