检索结果-内蒙古大学图书馆

Identifying catastrophic outlier photometric redshift estimates in the COSMOS field with machine learning methods

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Dennis, Mitchell T. Hu, Esther M. Cowie, Lennox L. Institute for Astronomy University of Hawai‘i at Mānoa 2680 Woodlawn Drive HonoluluHI96822 United States

We present the result of two binary classifier ensembled neural networks to identify catastrophic outliers for photo-z estimates within the COSMOS field utilizing only 8 and 5 photometric band passes, respectively. Our neural networks can correctly classify 55.6% and 33.3% of the true positives with few to no false positives. These methods can be used to reduce the errors caused by the errors in redshift estimates, particularly at high redshift. When applied to a larger data set with only photometric data available, our 8 band pass network increased the number of objects with a photo-z greater than 5 from 0.1% to 1.6%, and our 5 band pass network increased the number of objects with a photo-z greater than 5 from 0.2% to 1.8%. Copyright © 2025, The Authors. All rights reserved.

关键词： learning algorithms

Model based multi-agent reinforcement learning with tensor decompositions

学校读者我要写书评

暂无评论

arXiv 2021年

作者： van der Vaart, Pascal Mahajan, Anuj Whiteson, Shimon TU Delft University of Oxford

A challenge in multi-agent reinforcement learning is to be able to generalize over intractable state-action spaces. Inspired from Tesseract [Mahajan et al., 2021], this position paper investigates generalisation in state-action space over unexplored state-action pairs by modelling the transition and reward functions as tensors of low CP-rank. Initial experiments on synthetic MDPs show that using tensor decompositions in a model-based reinforcement learning algorithm can lead to much faster convergence if the true transition and reward functions are indeed of low rank. © 2021, CC BY.

关键词： learning algorithms

Vehicle behavior prediction and generalization using imbalanced learning techniques

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Westny, Theodor Frisk, Erik Olofsson, Björn The Division of Vehicular Systems Department of Electrical Engineering Linköping University Linköping Sweden

The use of learning-based methods for vehicle behavior prediction is a promising research topic. However, many publicly available data sets suffer from class distribution skews which limits learning performance if not addressed. This paper proposes an interaction-aware prediction model consisting of an LSTM autoencoder and SVM classifier. Additionally, an imbalanced learning technique, the multiclass balancing ensemble is proposed. Evaluations show that the method enhances model performance, resulting in improved classification accuracy. Good generalization properties of learned models are important and therefore a generalization study is done where models are evaluated on unseen traffic data with dissimilar traffic behavior stemming from different road configurations. This is realized by using two distinct highway traffic recordings, the publicly available NGSIM US-101 and I80 data sets. Moreover, methods for encoding structural and static features into the learning process for improved generalization are evaluated. The resulting methods show substantial improvements in classification as well as generalization performance. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

FedPAGE: A fast local stochastic gradient method for communication-efficient federated learning

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Zhao, Haoyu Li, Zhize Richtárik, Peter Princeton University United States Kaust Saudi Arabia

Federated Averaging (FedAvg, also known as Local-SGD) (McMahan et al., 2017) is a classical federated learning algorithm in which clients run multiple local SGD steps before communicating their update to an orchestrating server. We propose a new federated learning algorithm, FedPAGE, able to further reduce the communication complexity by utilizing the recent optimal PAGE method (Li et al., 2021) instead of plain SGD in FedAvg. We show that FedPAGE uses much fewer communication rounds than previous local methods for both federated convex and nonconvex optimization. Concretely, 1) in the convex setting, the number of communication rounds of FedPAGE is O(N3/4/Sϵ), improving the best-known result O(N/Sϵ) of SCAFFOLD (Karimireddy et al., 2020) by a factor of N1/4, where N is the total number of clients (usually is very large in federated learning), S is the sampled subset of clients in each communication round, and ϵ is the target error;2) in the nonconvex setting, the number of communication rounds of FedPAGE is O(√ N+S/Sϵ2), improving the best-known result O(N2/3/S2/3ϵ2) of SCAFFOLD (Karimireddy et al., 2020) by a factor of N1/6S1/3, if the sampled clients S ≤ √N. Note that in both settings, the communication cost for each round is the same for both FedPAGE and SCAFFOLD. As a result, FedPAGE achieves new state-of-the-art results in terms of communication complexity for both federated convex and nonconvex optimization. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

Meta-learning an Inference Algorithm for Probabilistic Programs

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Che, Gwonsoo Yang, Hongseok KAIST Daejeon Korea Republic of

We present a meta-algorithm for learning a posterior-inference algorithm for restricted probabilistic programs. Our meta-algorithm takes a training set of probabilistic programs that describe models with observations, and attempts to learn an efficient method for inferring the posterior of a similar program. A key feature of our approach is the use of what we call a white-box inference algorithm that extracts information directly from model descriptions themselves, given as programs. Concretely, our white-box inference algorithm is equipped with multiple neural networks, one for each type of atomic command, and computes an approximate posterior of a given probabilistic program by analysing individual atomic commands in the program using these networks. The parameters of the networks are learnt from a training set by our meta-algorithm. We empirically demonstrate that the learnt inference algorithm generalises well to programs that are new in terms of both parameters and model structures, and report cases where our approach achieves greater test-time efficiency than alternative approaches such as HMC. The overall results show the promise as well as remaining challenges of our approach. Copyright © 2021, The Authors. All rights reserved.

关键词： learning algorithms

Unsupervised learning of correlated quantum dynamics on disordered lattices

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Kenig, Miri Lahini, Yoav Department of Condensed Matter Physics School of Physics and Astronomy Tel-Aviv University Tel Aviv6997801 Israel

Quantum particles co-propagating on disordered lattices develop complex non-classical correlations due to an interplay between quantum statistics, inter-particle interactions, and disorder. Here we present a deep learning algorithm based on Generative Adversarial Networks, capable of learning these correlations and identifying the physical control parameters in a completely unsupervised manner. After one-time training on a data set of unlabeled examples, the algorithm can generate, without further calculations, a much larger number of unseen yet physically correct new examples. Furthermore, the knowledge distilled in the algorithm's latent space identifies disorder as the relevant control parameter. This allows post-training tuning of the level of disorder in the generated samples to values the algorithm was not explicitly trained on. Finally, we show that a trained network can accelerate the learning of new, more complex problems. These results demonstrate the ability of neural networks to learn the rules of correlated quantum dynamics in an unsupervised manner and offer a route to their use in quantum simulations and computation. © 2021, CC BY.

关键词： learning algorithms

META DISCOVERY: learning TO DISCOVER NOVEL CLASSES GIVEN VERY LIMITED DATA

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Chi, Haoang Liu, Feng Han, Bo Yang, Wenjing Lan, Long Liu, Tongliang Niu, Gang Zhou, Mingyuan Sugiyama, Masashi National University of Defense Technology China Hong Kong Baptist University Hong Kong University of Technology Sydney Australia The University of Sydney Australia RIKEN AIP Japan The University of Texas Austin United States The University of Tokyo Japan

In novel class discovery (NCD), we are given labeled data from seen classes and unlabeled data from unseen classes, and we train clustering models for the unseen classes. However, the implicit assumptions behind NCD are still unclear. In this paper, we demystify assumptions behind NCD and find that high-level semantic features should be shared among the seen and unseen classes. Based on this finding, NCD is theoretically solvable under certain assumptions and can be naturally linked to meta-learning that has exactly the same assumption as NCD. Thus, we can empirically solve the NCD problem by meta-learning algorithms after slight modifications. This meta-learning-based methodology significantly reduces the amount of unlabeled data needed for training and makes it more practical, as demonstrated in experiments. The use of very limited data is also justified by the application scenario of NCD: since it is unnatural to label only seen-class data, NCD is sampling instead of labeling in causality. Therefore, unseen-class data should be collected on the way of collecting seen-class data, which is why they are novel and first need to be clustered. © 2021, CC BY.

关键词： learning algorithms

Batch active learning at scale

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Citovsky, Gui DeSalvo, Giulia Gentile, Claudio Karydas, Lazaros Rajagopalan, Anand Rostamizadeh, Afshin Kumar, Sanjiv Google Research

The ability to train complex and highly effective models often requires an abundance of training data, which can easily become a bottleneck in cost, time, and computational resources. Batch active learning, which adaptively issues batched queries to a labeling oracle, is a common approach for addressing this problem. The practical benefits of batch sampling come with the downside of less adaptivity and the risk of sampling redundant examples within a batch - a risk that grows with the batch size. In this work, we analyze an efficient active learning algorithm, which focuses on the large batch setting. In particular, we show that our sampling method, which combines notions of uncertainty and diversity, easily scales to batch sizes (100K-1M) several orders of magnitude larger than used in previous studies and provides significant improvements in model training efficiency compared to recent baselines. Finally, we provide an initial theoretical analysis, proving label complexity guarantees for a related sampling method, which we show is approximately equivalent to our sampling method in specific settings. © 2021, CC BY-NC-SA.

关键词： learning algorithms

Backpropagation and the brain

学校读者我要写书评

暂无评论

NATURE REVIEWS NEUROSCIENCE 2020年第6期21卷 335-346页

作者： Lillicrap, Timothy P. Santoro, Adam Marris, Luke Akerman, Colin J. Hinton, Geoffrey DeepMind London England UCL Ctr Computat Math & Phys London England Univ Oxford Dept Pharmacol Oxford England Univ Toronto Dept Comp Sci Toronto ON Canada Google Brain Toronto ON Canada

During learning, the brain modifies synapses to improve behaviour. In the cortex, synapses are embedded within multilayered networks, making it difficult to determine the effect of an individual synaptic modification on the behaviour of the system. The backpropagation algorithm solves this problem in deep artificial neural networks, but historically it has been viewed as biologically problematic. Nonetheless, recent developments in neuroscience and the successes of artificial neural networks have reinvigorated interest in whether backpropagation offers insights for understanding learning in the cortex. The backpropagation algorithm learns quickly by computing synaptic updates using feedback connections to deliver error signals. Although feedback connections are ubiquitous in the cortex, it is difficult to see how they could deliver the error signals required by strict formulations of backpropagation. Here we build on past and recent developments to argue that feedback connections may instead induce neural activities whose differences can be used to locally approximate these signals and hence drive effective learning in deep networks in the brain. The backpropagation of error (backprop) algorithm is frequently used to train deep neural networks in machine learning, but it has not been viewed as being implemented by the brain. In this Perspective, however, Lillicrap and colleagues argue that the key principles underlying backprop may indeed have a role in brain function.

关键词： Cortex learning algorithms Long-term potentiation Network models Neurophysiology