Avatar

Eric Fischer

Statistics Ph.D. student, Generative Modeling, Representation Learning

University of California, Los Angeles

About Me

I’m a Ph.D. student from the Department of Statistics at the University of California, Los Angeles. I research generative models and representation learning for language at the Center for Vision, Cognition, Learning, and Autonomy at UCLA under my advisor Dr. Ying Nian Wu. Here is my Github.

I earned a M.S. from the Department of Computer Science at UCLA, submitting my thesis “Deep Generative Classifier with Short Run Inference.” In this work, a deep generative classifier uses Short Run Markov Chain Monte Carlo inference, Langevin dynamics, and backpropagation through time to achieve similar classification accuracy to an analogous discriminative classifier, while having the advantages that it can generate data, learn unsupervised with additional unlabeled data, and exhibit robustness to adversarial attacks due to the stochasticity of the Langevin equation and the top-down architecture of the underlying generator network.

Before my M.S., I worked as a Full Stack Software Engineer in the San Francisco bay area for three years, most recently at NatureBox in Redwood City. Before this, I earned a Bachelor of Arts from the Department of Philosophy at UCLA, focusing on the philosophy of language and propositional and first-order logic.

Education

  • Ph.D. Statistics, Present

    University of California, Los Angeles

  • M.S. Computer Science with Thesis, 2020

    University of California, Los Angeles

  • B.A. Philosophy, 2013

    University of California, Los Angeles

Publications

Deep Generative Classifier with Short Run Inference

Deep generative classifier uses Short Run Markov Chain Monte Carlo inference, Langevin dynamics, and backpropagation through time to …

Learning Multi-Layer Latent Variable Model via Variational Optimization of Short Run MCMC for Approximate Inference

Short-run MCMC residual network outperforms a variational autoencoder in regard to image reconstruction error and image synthesis …

Statistical Models for Marr’s Paradigm

Credited as contributing author; Main contributor to the Preface, Introduction, and Chapters 1 and 2 of this book authored by Dr. …

Stochastic Grammars for Scene Parsing

Edited several chapters of this unpublished book authored by Dr. Song-Chun Zhu and my advisor Dr. Ying Nian Wu

Research

Exact Sampling with Coupled Markov Chains and Swendsen-Wang Cluster Sampling of the Ising Model

Convergence analysis of exact sampling with Gibbs sampler and coupled Markov chains vs. cluster sampling with Swendsen-Wang algorithm

Topology Adaptive Snakes Improve Mask Generation for Image Inpainting

T-snake deformable models, which can segment complex-shaped structures, improve generated masks passed to GAN for image inpainting

Implementation and Convergence Analysis of First-Order Optimization Methods for a CNN

Convergence analysis and Python implementations of SGD, SGD with momentum, SGD with Nesterov momentum, RMSprop, and Adam optimizers

Formulation of Variational Lower Bound and Application of VAE to MNIST Dataset

Formulation of evidence lower bound (ELBO) for variational autoencoder and application to synthesizing binary images

Experience

 
 
 
 
 

Teaching Assistant

University of California, Los Angeles

Mar 2020 – Present Los Angeles, CA
Have served as a Teaching Assistant for many undergraduate and graduate statistics courses at UCLA
 
 
 
 
 

Graduate Researcher

Center for Vision, Cognition, Learning, and Autonomy

Dec 2018 – Present Los Angeles, CA
Carry out research in generative modeling and representation learning for language
 
 
 
 
 

Full Stack Software Engineer

NatureBox

Mar 2016 – Dec 2017 Redwood City, CA
Core contributor to new Flux/React web application created after company added direct-to-consumer business; Led various projects including a payment processor migration, addition of Amazon payments, and a 2nd version of API
 
 
 
 
 

Software Engineer

Cinemagram

Sep 2015 – Dec 2015 San Francisco, CA
Worked with Python, Ruby, and SQL code to construct internal data management interfaces and tools