Abstract: 

The drive towards Artificial Intelligence (AI) has been an effort since the 1950's. From attempts to develop intelligent search algorithms to attempts to model computer intelligence by systematic Knowledge Representation (KR) of the world's data, in the 1980's. These efforts lost excitement for a few reasons including expensive, slow hardware, and limited data. Deep Learning is a subset of Machine Learning (ML) -- which is a subset of AI -- that extends the concept of Artificial Neural Networks (ANN) to uncover hidden patterns in unstructured datasets. Due to the current ubiquity of data (Big Data), and availability of on-demand, inexpensive, and parallel hardware such as Graphics Processing Units (GPUs) on Amazon EC2, Deep Learning has revitalized the excitement in AI. Breakthrough results can be seen in industry applications such as robotics, computer vision, healthcare, cyber security, text and voice conversational interfaces. Apache MXNet is a scalable, and multi-language ML engine to ease the development of deep neural networks. It blends declarative symbolic expression with imperative tensor computation, thus offering computation and memory efficiency. MXNet models are natively portable to various heterogeneous systems, ranging from mobile devices to distributed cloud-hosted GPU clusters. In this workshop, we work through the pipeline of provisioning, training and deploying deep learning applications using Apache MXNet on Amazon EC2. We cover applications such as image recognition and recommender systems. The participants will learn how to get up and running with the AWS Deep Learning Amazon Machine Image (AMI) in few minutes, and write a deep learning program in a few lines of codes in their favorite language (Python, Scala, R, or more). We will cover training models on one or multiple GPUs. Finally, we discuss design patterns for deploying MXNet deep learning applications in the cloud or on edge devices such as drones cars, or mobile phones.

Bio: 

Dan Mbanga is a Sr. Technical Account Manager at Amazon Web Services in New York. He focuses on Machine Learning and Big Data workloads. Dan helps AWS customers innovate on the cloud, in designing and building enterprise-scale analytics environments that satisfy requirements from different compliance regimes across industries. Prior to this, he built and led teams of specialized Big Data and DevOps support engineers at Amazon Development Center in Cape Town, South Africa. Before to joining Amazon, he lived and worked in several countries across South America, Africa, Europe and Asia; either delivering mobile payment software or building remote sites. Originally from Cameroon, Dan holds a BSc in Physics with minor in Computer Science, as well several industry certifications.