Trans. Mach. Learn. Res.最新文献_第2页

Revisiting adversarial training for the worst-performing class 对表现最差的班级重新进行对抗性训练

Trans. Mach. Learn. Res. Pub Date : 2023-02-17 DOI: 10.48550/arXiv.2302.08872

T. Pethick, Grigorios G. Chrysos, V. Cevher

引用次数: 3

On a continuous time model of gradient descent dynamics and instability in deep learning 深度学习中梯度下降动力学和不稳定性的连续时间模型

Trans. Mach. Learn. Res. Pub Date : 2023-02-03 DOI: 10.48550/arXiv.2302.01952

Mihaela Rosca, Yan Wu, Chongli Qin, B. Dherin

引用次数: 3

Dual PatchNorm 双重PatchNorm

Trans. Mach. Learn. Res. Pub Date : 2023-02-02 DOI: 10.48550/arXiv.2302.01327

Manoj Kumar, Mostafa Dehghani, N. Houlsby

引用次数: 3

Fourier Sensitivity and Regularization of Computer Vision Models 计算机视觉模型的傅里叶灵敏度和正则化

Trans. Mach. Learn. Res. Pub Date : 2023-01-31 DOI: 10.48550/arXiv.2301.13514

K. Krishnamachari, See-Kiong Ng, Chuan-Sheng Foo

{"title":"Fourier Sensitivity and Regularization of Computer Vision Models","authors":"K. Krishnamachari, See-Kiong Ng, Chuan-Sheng Foo","doi":"10.48550/arXiv.2301.13514","DOIUrl":"https://doi.org/10.48550/arXiv.2301.13514","url":null,"abstract":"Recent work has empirically shown that deep neural networks latch on to the Fourier statistics of training data and show increased sensitivity to Fourier-basis directions in the input. Understanding and modifying this Fourier-sensitivity of computer vision models may help improve their robustness. Hence, in this paper we study the frequency sensitivity characteristics of deep neural networks using a principled approach. We first propose a basis trick, proving that unitary transformations of the input-gradient of a function can be used to compute its gradient in the basis induced by the transformation. Using this result, we propose a general measure of any differentiable model's Fourier-sensitivity using the unitary Fourier-transform of its input-gradient. When applied to deep neural networks, we find that computer vision models are consistently sensitive to particular frequencies dependent on the dataset, training method and architecture. Based on this measure, we further propose a Fourier-regularization framework to modify the Fourier-sensitivities and frequency bias of models. Using our proposed regularizer-family, we demonstrate that deep neural networks obtain improved classification accuracy on robustness evaluations.","PeriodicalId":432739,"journal":{"name":"Trans. Mach. Learn. Res.","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-31","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128895624","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 1

PRUDEX-Compass: Towards Systematic Evaluation of Reinforcement Learning in Financial Markets PRUDEX-Compass:迈向金融市场强化学习的系统评估

Trans. Mach. Learn. Res. Pub Date : 2023-01-14 DOI: 10.48550/arXiv.2302.00586

Shuo Sun, Molei Qin, Xinrun Wang, Bo An

{"title":"PRUDEX-Compass: Towards Systematic Evaluation of Reinforcement Learning in Financial Markets","authors":"Shuo Sun, Molei Qin, Xinrun Wang, Bo An","doi":"10.48550/arXiv.2302.00586","DOIUrl":"https://doi.org/10.48550/arXiv.2302.00586","url":null,"abstract":"The financial markets, which involve more than $90 trillion market capitals, attract the attention of innumerable investors around the world. Recently, reinforcement learning in financial markets (FinRL) has emerged as a promising direction to train agents for making profitable investment decisions. However, the evaluation of most FinRL methods only focuses on profit-related measures and ignores many critical axes, which are far from satisfactory for financial practitioners to deploy these methods into real-world financial markets. Therefore, we introduce PRUDEX-Compass, which has 6 axes, i.e., Profitability, Risk-control, Universality, Diversity, rEliability, and eXplainability, with a total of 17 measures for a systematic evaluation. Specifically, i) we propose AlphaMix+ as a strong FinRL baseline, which leverages mixture-of-experts (MoE) and risk-sensitive approaches to make diversified risk-aware investment decisions, ii) we evaluate 8 FinRL methods in 4 long-term real-world datasets of influential financial markets to demonstrate the usage of our PRUDEX-Compass, iii) PRUDEX-Compass together with 4 real-world datasets, standard implementation of 8 FinRL methods and a portfolio management environment is released as public resources to facilitate the design and comparison of new FinRL methods. We hope that PRUDEX-Compass can not only shed light on future FinRL research to prevent untrustworthy results from stagnating FinRL into successful industry deployment but also provide a new challenging algorithm evaluation scenario for the reinforcement learning (RL) community.","PeriodicalId":432739,"journal":{"name":"Trans. Mach. Learn. Res.","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131781116","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Risk Sensitive Dead-end Identification in Safety-Critical Offline Reinforcement Learning 安全关键离线强化学习中的风险敏感死角识别

Trans. Mach. Learn. Res. Pub Date : 2023-01-13 DOI: 10.48550/arXiv.2301.05664

Taylor W. Killian, S. Parbhoo, M. Ghassemi

引用次数: 2

A Stochastic Proximal Polyak Step Size 随机近端Polyak步长

Trans. Mach. Learn. Res. Pub Date : 2023-01-12 DOI: 10.48550/arXiv.2301.04935

Fabian Schaipp, Robert Mansel Gower, M. Ulbrich

引用次数: 7

Exploring Efficient Few-shot Adaptation for Vision Transformers 探索视觉变形器的高效少镜头自适应

Trans. Mach. Learn. Res. Pub Date : 2023-01-06 DOI: 10.48550/arXiv.2301.02419

C. Xu

{"title":"Exploring Efficient Few-shot Adaptation for Vision Transformers","authors":"C. Xu","doi":"10.48550/arXiv.2301.02419","DOIUrl":"https://doi.org/10.48550/arXiv.2301.02419","url":null,"abstract":"The task of Few-shot Learning (FSL) aims to do the inference on novel categories containing only few labeled examples, with the help of knowledge learned from base categories containing abundant labeled training samples. While there are numerous works into FSL task, Vision Transformers (ViTs) have rarely been taken as the backbone to FSL with few trials focusing on naive finetuning of whole backbone or classification layer.} Essentially, despite ViTs have been shown to enjoy comparable or even better performance on other vision tasks, it is still very nontrivial to efficiently finetune the ViTs in real-world FSL scenarios. To this end, we propose a novel efficient Transformer Tuning (eTT) method that facilitates finetuning ViTs in the FSL tasks. The key novelties come from the newly presented Attentive Prefix Tuning (APT) and Domain Residual Adapter (DRA) for the task and backbone tuning, individually. Specifically, in APT, the prefix is projected to new key and value pairs that are attached to each self-attention layer to provide the model with task-specific information. Moreover, we design the DRA in the form of learnable offset vectors to handle the potential domain gaps between base and novel data. To ensure the APT would not deviate from the initial task-specific information much, we further propose a novel prototypical regularization, which maximizes the similarity between the projected distribution of prefix and initial prototypes, regularizing the update procedure. Our method receives outstanding performance on the challenging Meta-Dataset. We conduct extensive experiments to show the efficacy of our model.","PeriodicalId":432739,"journal":{"name":"Trans. Mach. Learn. Res.","volume":"66 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-01-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115320638","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 7

Reusable Options through Gradient-based Meta Learning 基于梯度元学习的可重用选项

Trans. Mach. Learn. Res. Pub Date : 2022-12-22 DOI: 10.48550/arXiv.2212.11726

David Kuric, H. V. Hoof

引用次数: 0

Bridging Graph Position Encodings for Transformers with Weighted Graph-Walking Automata 加权图行走自动机变压器桥图位置编码

Trans. Mach. Learn. Res. Pub Date : 2022-12-13 DOI: 10.48550/arXiv.2212.06898

Patrick M. Soga, David Chiang

引用次数: 0