A Review of Auto-Encoders

Thrilled to announce the very first meetup. Big thank you to SkillsMatter (https://skillsmatter.com) for hosting us and Persontyle (http://www.persontyle.com/) for sponsoring.

** To guarantee a place you must register with Skillsmatter as well (https://skillsmatter.com/meetups/6216-a-review-of-auto-encoders) **

--Dirk

@elazungu (https://twitter.com/elazungu)

• Auto-Encoders - Piotr Mirowski (http://twitter.com/mirowskipiotr)

We will review some basics of deep learning: learning sparse or compact representations using stacked auto-encoders, in an unsupervised or semi-supervised way (with partially labelled data) and using the stochastic gradient descent algorithm. In this hands-on tutorial, I will share simple Matlab code for training sparse code representations of handwritten digits and present applications to text categorization.

Bio: Piotr Mirowski is a data scientist who recently joined Microsoft as a software engineer at the London Bing offices where he currently works on query formulation. He is passionate about deep learning, statistical language models and time series analysis, passions he acquired by doing his PhD at New York University under the guidance of neural network guru Yann LeCun. Piotr loves to solve real world problems and has focused his studies on many different projects, including epileptic seizure prediction from EEG and inference of gene regulation networks. In his previous job as a research scientist at Bell Labs, Piotr delivered an electric load forecasting solution to a power utility, hacked a Kinect-powered mobile robot that mapped the WiFi signal strength (for indoor geo-localization) and coded Simultaneous Localization and Mapping from the pocket (reconstructing a 3D trajectory using only sensors present on a smartphone). He loves backpacking and improv theatre acting.