Variational Autoencoders For Image Generation by Dr. Steven Flores

NOTE: Please be sure to go to 700 EAST El Camino Real and NOT West El Camino Real

Abstract
An autoencoder is a machine learning algorithm that represents unlabeled high-dimensional data as points in a low-dimensional space. A variational autoencoder (VAE) is an autoencoder that represents unlabeled high-dimensional data as low-dimensional probability distributions. In addition to data compression, the randomness of the VAE algorithm gives it a second powerful feature: the ability to generate new data similar to its training data. For example, a VAE trained on images of faces can generate a compelling image of a new "fake" face. It can also map new features onto input data, such as glasses or a mustache onto the image of a face that initially lacks these features. In this talk, we will survey VAE model designs that use deep learning, and we will implement a basic VAE in TensorFlow. We will also demonstrate the encoding and generative capabilities of VAEs and discuss their industry applications.

Agenda:
• 18.30: Doors Open
• 18.50: RedisAI Lightning Talk by Dave Nielsen of Redis Labs
• 19.00: Main Presentation
• 19:45: Q/A
• 20:00: Mingling

Speaker
Steven Flores is a machine learning engineer and researcher at Comp Three Inc. He use deep learning and other cutting-edge methods in data modeling, computer vision, and NLP to find data-driven solutions that solve our clients' most difficult business challenges. Having worked in a variety of fields, spanning from computer science and AI to physics and mathematics, Steven blends expertise from different domains to design novel and practical solutions and then deploys them at scale. He earned his Ph.D in applied math from the University of Michigan in 2012 and is published in top research journals. While at Comp Three, Steven has worked with customers in real-estate appraisal, web search, and NLP.