JMLR 2020

162 papers
The 2020 edition of the Journal of Machine Learning Research includes work focused on algorithmic methodology, graph theory, and optimization. One paper by Frank Hutter and Marius Lindauer establishes best practices for scientific research in neural architecture search, addressing the need for standardized evaluation and reproducibility in automated machine learning. This emphasis on methodological rigor is a central theme among the year's contributions. Additional research explores theoretical aspects of machine learning, such as graph matching using partially-correct seeds and the properties of local minima in L1-minimization for complete dictionary learning. Authors including Bin Yu, Jiaming Xu, and Liren Yu contributed to these studies, which bridge the gap between practical implementation and fundamental mathematical analysis. The collected works represent a focus on refining existing search techniques and understanding the underlying landscapes of optimization problems.

Top Research Directions

Research directions most represented by JMLR 2020 papers.

Variational Inference as an Optimization Problem 3
Overcoming the Convexity Constraint in Modern Optimization 3
Proving Statistical Consistency for Complex Estimators 2
Learning and Estimation Under Model Misspecification 2
Inferring Latent Networks from Diffusion Timestamps 2
Handling Mixed Discrete and Continuous Data Types 2
How Network Topology Governs Decentralized Learning Convergence 2
Determinantal Point Processes for Diverse Subset Selection 2
Variance Underestimation in Mean-Field Variational Inference 2
Algorithmic and Hardware Strategies for Scalable Bayesian Inference 2
Alternating Optimization for Sparse Dictionary Learning 2
Geometry Aware Preconditioning for High Dimensional Optimization 2
Mitigating Parameter Explosion in High-Dimensional Time Series 2
Unified Evaluation Frameworks for Reproducible Machine Learning 2
Breaking the Curse of Horizon in Off Policy Evaluation 2

Top Authors

AuthorPapers
George Michailidis3
Aryan Mokhtari2
Bernhard Schölkopf2
Junhong Lin2
Kamil Ciosek2
Kun Zhang2
Michael A Osborne2
Mladen Kolar2
Nathan Kallus2
Ping Li2
Volkan Cevher2
Xiaotong Yuan2
Abhradeep Guha Thakurta1
Abir De1
Abla Kammoun1
Adam Golinski1
Adam Roberts1
Adam Smith1
Akshay Krishnamurthy1
Alberto Maria Metelli1

Impactful Papers

Papers from JMLR 2020 that shaped research clusters.

Foundational

Influential

Papers

Title
The Maximum Separation Subspace in Sufficient Dimension Reduction with Categorical Response
The Optimal Ridge Penalty for Real-world High-dimensional Data Can Be Zero or Negative due to the Implicit Ridge Regularization
ThunderGBM: Fast GBDTs and Random Forests on GPUs
Trust-Region Variational Inference with Gaussian Mixture Models
Tslearn, A Machine Learning Toolkit for Time Series Data
Tuning Hyperparameters without Grad Students: Scalable and Robust Bayesian Optimisation with Dragonfly
Two-Stage Approach to Multivariate Linear Regression with Sparsely Mismatched Data
Unique Sharp Local Minimum in L1-minimization Complete Dictionary Learning
Variational Inference for Computational Imaging Inverse Problems
Weighted Message Passing and Minimum Energy Flow for Heterogeneous Stochastic Block Models with Side Information
Wide Neural Networks with Bottlenecks are Deep Gaussian Processes
WONDER: Weighted One-shot Distributed Ridge Regression in High Dimensions

« prev Page 4/4