Top 10 blog.evjang.com Competitors

bjlkeng.io

A short post on showing what happens if you keep summarizing a piece of text.

sassafras13.github.io

We are back again to revisit another piece of the MolGAN paper and this time we are looking at how they trained their GAN to generate molecules that satisfied particular design criteria [1]. If you recall from the overview of the MolGAN work, they designed a GAN that could generate novel molecules which met some design criteria, like being water soluble or druglike [1]. In order to train the GAN to meet these criteria, De Cao and Kipf used tools from reinforcement learning to optimize their generator [1]. Specifically, they used a policy gradient called Deep Deterministic Policy Gradient (DDPG) [1]. It has been a long time since I have read research related to reinforcement learning, so this post is going to cover some of the basics of reinforcement learning (RL) and introduce the role that policy gradients have in RL. We will introduce algorithms that solve for the policy gradient, such as REINFORCE. In a follow-on post, we will dig into DDPG itself and explain how it works in the context of MolGAN.

weideng.org

wei deng is a machine learning researcher and applied mathematician.

zacharyespiritu.com

Introduction

n-young.me

nick young's personal website

zzzdavid.tech

My experience attending and presenting at CVPR22

davidstutz.de

This week I was honored to speak at the Machine Learning Security Seminar organized by the Pattern Recognition and Applications Lab at University of Cagliari. I presented my work on relating adversarial robustness to flatness in the robust loss landscape, also touching on the relationship to weight robustness. In this article, I want to share the recording and slides of this talk.

gebob19.github.io

In this brief tutorial/example, I explain the best way to store videos in TfRecords for more efficient and faster model training in TensorFlow version 1 / 1.15.0 .