You are here

Bots & AI: Teaching Machines to See & Creating Intelligent Machine Agents



Thought Works
99 Madison Ave., 15th Floor
New York, NY
United States


Thursday, August 3, 2017 -
6:00pm to 9:00pm


This month, we will learn about how to teach machines to see and how to build a bot to handle all those annoying times while on customer service hold. 

Ryan Compton (Head of Applied Machine Learning at Clarifai) shares how to create a visual artificial intelligent (AI) system that gives machines the ability to identify and classify images.  

Jeff Smith (Founder of John Done) demonstrates how to build intelligent support agents bots that can ask questions on your behalf whether to humans or IVR phone trees -- no more call waiting or mind-numbing phone menus! 

This event is hosted by the meetup group Bots and Artificial Intelligence.


6:00pm - Networking and Pizza

6:30pm - Kickoff

6:35pm - Bot Demo: Yaniv Nissim of GoParrot

6:40pm - Talk #1: Jeff Smith on Intelligent Phone Agents

7:15pm - Talk #2: Ryan Compton on Teaching Machines to See

8:00pm - Extended Q&A and Open Member Discussion

Talk Descriptions:

Talk #1: 

Composition and Collaboration: Intelligent Agents Today by Jeff Smith This talk will discuss how we build conversational intelligent agents today, through the composition of skills and datasets, often from different organizations. We'll take a detailed look at all of the components of an intelligent agent, discussing what can be acquired from the outside versus what might be core to your organizational mission. We'll try to tease apart what is easy and commodity versus rare and valuable in the building of a machine learning system. Then, we'll zoom out to discuss this phenomenon at an industry level, exploring what this level of cross-organizational technical collaboration means for the future of intelligent agent development. Although some of these issues are very general, this talk will focus on conversational agents, addressable via voice and/or text. Bios: Jeff Smith is the co-founder of John Done, a startup building a voice assistant that manages your phone calls. For the past decade, he has been working on data science applications at various startups in New York, San Francisco, and Hong Kong. He’s a frequent speaker and blogger, and the author of Reactive Machine Learning Systems, an upcoming book from Manning on how to build real-world machine learning systems.

Talk #2: 

Teaching Machines to See Building datasets for machine learning has been a major problem in computer vision for over two decades now, and, because of its rich history and straightforward goal, serves as a great example of how the field has evolved. We'll illustrate how training modern convolutional neural networks (convnets) differs from research done in the past. In particular, we'll discuss the notion of a deconvolutional network and demonstrate how they can be used to visualize intermediate feature layers and the operation of a classifier. Ryan Compton currently heads applied machine learning at Clarifai. His day-to-day is designing datasets to train neural nets and then shipping them into production. Ryan holds a Ph.D. in mathematics from UCLA with a focus on sparsity-promoting optimization and was previously on staff at Howard Hughes laboratories. Some of his research has been covered in Forbes, the New York Observer, and Business Insider among other places. Bot Demo and Lessons Learned: Yaniv Nissim will demo and discuss GoParrot, an AI platform for food ordering with smart analytics and retention.