Events Add an event Speakers Talks Collections
 
MLconf Online 2020
November 6, 2020, Online
MLconf Online 2020
Request Q&A
MLconf Online 2020
From the conference
MLconf Online 2020
Request Q&A
Video
ML for the Real World: Distributionally Robust Extrapolation
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Add to favorites
100
I like 0
I dislike 0
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
  • Description
  • Transcript
  • Discussion

About the talk

The unprecedented prediction accuracy of modern machine learning beckons for its application in a wide range of real-world applications, including autonomous robots, medical decision-making, scientific experimental design, and many others.  A key challenge in such real-world applications is that the test cases are not well represented by the pre-collected training data.  To properly leverage learning in such domains, especially safety-critical ones, we must go beyond the conventional learning paradigm of maximizing average prediction accuracy with generalization guarantees that rely on strong distributional relationships between training and test examples.

In this talk, I will describe a distributionally robust learning framework that offers rigorous extrapolation guarantees under data distribution shift. This framework yields appropriately conservative yet still accurate predictions to guide real-world decision-making and is easily integrated with modern deep learning.  I will showcase the practicality of this framework in an application on agile robotic control.  I will also introduce a survey of other real-world applications that would benefit from this framework for future work.

About speaker

Anqi Liu
Machine Learning Researcher and Postdoctoral Scholar at Caltech Department of Computing and Mathematical Sciences

Anqi (Angie) Liu is a machine learning researcher and postdoctoral scholar at the Department of Computing and Mathematical Sciences in Caltech. She obtained her Ph.D. from the Department of Computer Science of the University of Illinois at Chicago. She works on distributionally robust learning, distribution shift, and interactive machine learning. She is interested in machine learning for safety-critical tasks and the social impact of AI and aims to design learning methods for more reliable systems in the real world.

View the profile
Share

Hey everyone. Thank you so much for joining us back from the break. I hope you're able to stretch your legs and get some coffee. I'm Paul McLaughlin on the presenting. Our next speaker is and either you supposed to do research. And I will give you a 10 minute warning and a 5 minute warning. I just see you. I have until Thanksgiving. Everyone in the corporation was amazing. No, that's what are some examples of real-world machine tasks. so, I'll just make

Uncle control, ask for money for and also very important application. What is it in. What's the kind of? How are you doing today? All? Emotionally mother? So you don't even want to make sure I see how the real world it's working when there's no rain. Marriott uses model for dancing because there's action. No, expiration can be added as a domestic problem. Example, if we want to go to the stupid for what makes me very in your face and it was. So this is the problem. So what will it be? The hottest Revelations dragon in this

problem? We have some food. So, what should be the population? Hard to hear your voice. I don't know if it's just me, but maybe others need to having a hard time hearing your voice. What movie is it? Well. So in the end is not in the right hand side is examples. Tell you that. How to be conservative and then probably too much from my. So, is nnn what useful for the state of emergency. So go, he is trying to step by step. So, they could not have to do a special relationship.

Example of how can we get to do some crappie schools from anniversary activation? The first, some background information. What's concerning our anniversary? The same thing? Let's see. What happens on the dyskinesia. The honest truth, Chef still have this game between some Lego stores. What's the temperature? So I'm going to have to we'll have this kind of the difference from our time difference between this model and the Furious 1. I think the problem here is very

Call Matthew Richardson reason. Go to Safeway. Buy. We give a prediction on how far away from my music. So, it was a prior knowledge that I still want otherwise dismal. Let's see how it works in the real world. What type of heater is the control with just fine and also being safe and the green line inside? And then that time but why? So So what's machines to extrapolate and office space in your safe, place. So, every time they said about Stacy. You are a modest operation.

So he has a very standard on control. Replikaz, how it works independently. Indeed.com. The Drone can also with Will my Thurman? This problem is cause it's also so important. So close to the optical. Aaron Sanchez. Why do some analysis Chestnut call me? The number of operation? So, as a summarization model. and then we As I was like, I want to know. And the other part of my have fundamental models that can keep think you and I can save. Great. Thank you so much for a fascinating presentation. You have any questions are for

our speaker. Please put them in the chat and I will happily read them for i n g. If not, I might ask you a question actually based on your hypothetical around the airplane. What would want to see in place before you boarded a plane? Seems like Guided by Machine morning and headed to build them a hypothetical you ask. Moving a horse that just says, is there an architectural difference between your model and Dan models? But then the difference is really trying to find this

word for this finding is well, first we are fine. Just trying to make sure the difference. Function with this song right to on time. Thank you so much and you for your presentation?

Cackle comments for the website

Buy this talk

Access to the talk “ML for the Real World: Distributionally Robust Extrapolation”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free

Ticket

Get access to all videos “MLconf Online 2020”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Ticket

Interested in topic “Artificial Intelligence and Machine Learning”?

You might be interested in videos from this event

February 4 - 5, 2021
Online
26
104
ai, application, bot, chatbot, conversation, data, design, healthcare, ml

Similar talks

Tianshi Gao
Principal AI Scientist at Cruise
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Nitin Sharma
Senior Research Scientist at PayPal Risk Sciences
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Meghana Ravikumar
Machine Learning Engineer at SigOpt
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free

Buy this video

Video
Access to the talk “ML for the Real World: Distributionally Robust Extrapolation”
Available
In cart
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free
Free

Conference Cast

With ConferenceCast.tv, you get access to our library of the world's best conference talks.

Conference Cast
949 conferences
37757 speakers
14408 hours of content