Picture for Matthew J. Johnson

Matthew J. Johnson

Decomposing reverse-mode automatic differentiation

May 20, 2021
Roy Frostig, Matthew J. Johnson, Dougal Maclaurin, Adam Paszke, Alexey Radul

* Presented at the LAFI 2021 workshop at POPL, 17 January 2021

Access Paper or Ask Questions

SOLAR: Deep Structured Representations for Model-Based Reinforcement Learning

Feb 20, 2019
Marvin Zhang, Sharad Vikram, Laura Smith, Pieter Abbeel, Matthew J. Johnson, Sergey Levine

Figure 1 for SOLAR: Deep Structured Representations for Model-Based Reinforcement Learning

Figure 2 for SOLAR: Deep Structured Representations for Model-Based Reinforcement Learning

Figure 3 for SOLAR: Deep Structured Representations for Model-Based Reinforcement Learning

Figure 4 for SOLAR: Deep Structured Representations for Model-Based Reinforcement Learning

* under review for ICML 2019

Access Paper or Ask Questions

Autoconj: Recognizing and Exploiting Conjugacy Without a Domain-Specific Language

Nov 29, 2018
Matthew D. Hoffman, Matthew J. Johnson, Dustin Tran

Figure 1 for Autoconj: Recognizing and Exploiting Conjugacy Without a Domain-Specific Language

Figure 2 for Autoconj: Recognizing and Exploiting Conjugacy Without a Domain-Specific Language

Figure 3 for Autoconj: Recognizing and Exploiting Conjugacy Without a Domain-Specific Language

* Appears in Neural Information Processing Systems, 2018. Code available at https://github.com/google-research/autoconj

Access Paper or Ask Questions

Composing graphical models with neural networks for structured representations and fast inference

Jul 07, 2017
Matthew J. Johnson, David Duvenaud, Alexander B. Wiltschko, Sandeep R. Datta, Ryan P. Adams

Figure 1 for Composing graphical models with neural networks for structured representations and fast inference

Figure 2 for Composing graphical models with neural networks for structured representations and fast inference

Figure 3 for Composing graphical models with neural networks for structured representations and fast inference

Figure 4 for Composing graphical models with neural networks for structured representations and fast inference

* v5 fixes tex compilation bugs and also a math bug in the statement and proof of Prop. 4.1 (and D.3). v4 adds two paragraphs to the related work section and fixes typos in the appendices. v3 fixes some typos in the appendices. v2 is a rewrite from v1 to be more readable and to include detailed appendices

Access Paper or Ask Questions

Recurrent switching linear dynamical systems

Oct 26, 2016
Scott W. Linderman, Andrew C. Miller, Ryan P. Adams, David M. Blei, Liam Paninski, Matthew J. Johnson

Figure 1 for Recurrent switching linear dynamical systems

Figure 2 for Recurrent switching linear dynamical systems

Figure 3 for Recurrent switching linear dynamical systems

Figure 4 for Recurrent switching linear dynamical systems

* 15 pages, 6 figures

Access Paper or Ask Questions