Class Central is learner-supported. When you buy through links on our site, we may earn an affiliate commission.

Stanford University

Stanford Seminar - Building the Smartest and Open Virtual Assistant to Protect Privacy - Monica Lam

Stanford University via YouTube

Overview

Save Big on Coursera Plus. 7,000+ courses at $160 off. Limited Time Only!

Virtual assistants, providing a voice interface to web services and IoTs, can potentially develop into monopolistic platforms that threaten consumer privacy and open competition. This talk presents Almond as an open-source alternative.

We introduce a more cost-effective and robust methodology and tool set for building virtual assistants. We use an end-to-end neural model that translates natural language to directly executable code in our new virtual assistant programming language, called ThingTalk. As annotating real data training is prohibitively expensive, we train our neural network with mostly data synthesized with our Genie data engineering tool. Genie can automatically generate a dialogue agent from a database in given domains. In addition, it is designed to be extensible: developers can supply new domain schemas and APIs, domain-independent sentence templates and dialogue state machines in any language, as well as new ThingTalk constructs. Using this methodology, Almond has been shown to answer complex questions more accurately than existing commercial assistants in the domain of restaurant reservations.

To protect privacy, our Almond virtual assistant prototype has a federated design that lets users own their data and share their digital assets with full control. Almond has been integrated into Home Assistant, an open-source local gateway that can connect to hundreds of different IoT devices. By making all the tools, neural models, training data open-source, we wish to support open-world collaboration to create the smartest and open virtual assistant. Almond and more information are available on http://oval.cs.stanford.edu.

This research is supported by the NSF under grant No. 1900638.

Syllabus

Introduction.
Consumer Privacy.
Virtual Assistant Oligopoly.
Threats of Virtual Assistants.
Technical Barrier to Entry.
Scaling to the Future of Virtual Assistants.
This Talk.
Alexa: Syntax-Dependent Representation.
Alexa's 2-Step Approach.
Idea 1: End-to-End Translation.
Unique Semantic Representation.
Idea 2: Training-Data Engineering.
Genie: Synthesizes question/code from a schema.
Semantic Parser Generation Thingpedia.
Query Comparison - 4 Platforms.
Comparison of 415 Restaurant Questions.
Today's Dialogue Trees: Laborious & Brittle.
Annotate 1 Dialogue at a Time.
Genie: Transactional Dialogues State Machine.
Modular, Reusable Technical Stack.
Contextual Language Understanding Neural Model.
Preliminary Results.
Zero-Shot Learning Result.
Three Research Results.
Protect Privacy with an Open Federated Architecture.
Example: Asthma Patient.

Taught by

Stanford Online

Reviews

5.0 rating, based on 1 Class Central review

Start your review of Stanford Seminar - Building the Smartest and Open Virtual Assistant to Protect Privacy - Monica Lam

  • SINGEPOGU KUMAR
    very informative and understanding way of explanation. I have many doubts regarding this topic before start of the course now im very clear of it now. I was delighted with it and look forward to coming back for more classes in the future. I loved the fact that even though I was in a class with other people, it was very csatisfying for me. I love all the classes I have taken through. i think the instructors are very experienced in the programs they are teaching. I have recommended this site to all my friends. I look forward to taking more classes

Never Stop Learning.

Get personalized course recommendations, track subjects and courses with reminders, and more.

Someone learning on their laptop while sitting on the floor.