Shattered Gradients: A Weekly Deep Learning Review

Krithik Ramesh, a student ML engineer/polymath from Colorado, joins Brendon to talk about ML as it relates to the arts, philosophy, automation, politics, and the future. Please send comments to .

Brendon interviews Kevin Meng, a student ML researcher from Plano, Texas, about his research on using machine learning, specifically many-to-many recurrent models, to reconstruct a person's pose using RF data even when a wall or obstruction is present, taking advantage of time series data. Kevin also shares insights on presenting ML research to the public, and his plans for future research directions relating to natural language processing for media fact checking! Please send comments to .

Brendon and Anish interview Filip Ku?erak, a student from Slovakia, about his research on using genetic algorithms to generate realistic, lifelike trees in a simulated environment. We discuss how one develops a realistic simulation that can be optimized using genetic algorithms, the concepts that enable genetic algorithms to work, and how they relate to other machine learning methods. Please send comments to .

Brendon and Anish interview Zach Trefler, a student at the University of Toronto, about his research on designing neural networks to create adversarial examples to defeat a voiceprinting system. We discuss how generative-adversarial networks are applied and how they can be applied to generate adversarial examples, in order to trick a security system into classifying the examples as a completely different user. Please send comments to .

Brendon and Anish interview Max von Wolff, a student from Mayen, Germany, about his research on short-term predictions of storm movement using deep learning with a network of distributed weather observation devices. We discuss the advantages and difficulties of processing data collected with embedded devices in the field, the use of machine learning methods such as autoencoders for processing this data, and Max's plans to scale up his research! Please send comments to .

Jacky Zhao, a student and deep learning researcher in Vancouver, joins Anish and Brendon to discuss ethical challenges and solutions in the world of deep learning. We talk about examples of ethical AI issues in industry, as well as Jacky's unique research applying deep learning to help people with hearing impairments. Please send comments to .

David Holz, founder and CTO of Leap Motion, interviews with Brendon and Anish on building a career in Silicon Valley and the future of deep learning, live at Intel ISEF 2019 in Phoenix, Arizona. Holz offers insightful commentary on the beginnings of his own career, including Intel ISEF, the world of research, and the founding and early days of Leap Motion. He goes on to discuss advice for students looking to start a career in deep learning, the most promising areas of the rapidly growing field, and some future directions it may take. Many thanks to Mr. Holz for joining us to share his experience and insights! Please send comments to .

Anish and Brendon open our first episode by talking about the technologies behind artificial intelligence: What is deep learning? How does it work? What key problems can it solve? We continue to discuss a few specific applications of deep learning in autonomous driving, for both indoor and outdoor vehicles. Anish discusses semantic segmentation (a powerful technology for object recognition in images), and we wrap up the episode with Q/A. Please send comments, concerns, suggestions, compliments, insults, and requests to join the show as a guest to .