Causality 101 with Robert Ness


I am on the line with Robert Osa Zoa Ness Robert is a machine learning research engineer at Gammel on and an instructor at northeastern North Eastern University Robert and I met at the last nerves conference where he had an accepted poster session around his paper integrating hitting Markov processes with structural. Causal Modeling Enables counterfactual inference in complex systems which he also presented at the black doc in a in a workshop This kicked off a bunch of great conversations between the two of US leading ultimately to collaboration. That we'll talk a little bit about in this conversation. Robert thanks so much for joining me on the Tuomo. Ai podcast thanks for having me Sam. You're injured us. It makes me think I should've. It came up with more clever name for that paper. You know what a lot of papers we talk about on. This show are quite the mouthful so yours is no exception exception Maybe someone will build a model. You know that. seeks to determine a inverse correlation or correlation between the lengthiness papers the title and It's number of citations or something like that. Let's set that aside for now and have you spent a few minutes introducing yourself. How did you get started in machine learning what piqued your interest You know ultimately will be spending a lot of time here talking about causality. How did you come to Become interested in that you know my path to machine learning was a bit. I'd say unconventional I started off working In Asia Tanna specifically I was the degree at Hopkins in International Studies and was planning adding to pursue a degree in economic In economics focusing on economic development I got involved with some Internet companies out in Beijing That got me into coding. And database is in data in general and I decided I was interested in in debt in that and went to apply for programs in statistics. Particularly with a focus on computational statistics I back to the states came back to the states went to Purdue University to do my PhD in stats My PhD work was on causal. Inference graphical models Basically how to learn causal models from data particularly in the context of systems systems biology and from then after I graduated I went to trade industry. Got It now. We hear very frequently folks refer to their path into machine learning as unconventional are indirect In your case you came into an interesting gaming net leads you to apply live for Or into Grad School for statistic. What was that particular connection really? It's when you're on the back end of an APP and you're looking at the data and you're realizing that there's a lot of insights to be had if only we could model this data and turn it into some service on the front end Um I realized I mean this was you know people had were just kinda starting to talk about data science and then Hell Varian had just recently came out and said I said that's the districts is the new sexiest will. I can't remember the exact quote was pick your Metaphor Metaphor New Black statistics is the new. I don't know Rockstar and so And Yeah that's that's kind of why pivoted to do the two stats in machine. Learning I guess through stats view. May people might argue whether or not stats machine. Learning Are same thing. Might the problems that I was working on my PhD or using Publicity graphical models so which has strong roots in artificial intelligence. So that was my introduction machine learning. Yeah one of the things that's come up in our conversations about causality and The work that you're doing with your courses is the idea that it historically talking about causality has been the you know the domain of statisticians and in Yeah folks like economists And that a lot of that conversation is inaccessible or isn't really tailored to do the needs of developers and data scientists machine learning engineers. I didn't realize all the time we were talking about that. That your background wasn't economics. You you have some of the exposure to the way that causality is has been traditionally kind of us and talked about. Maybe I guess I'll just use this as a segue to Kind of opening up the floor to to ask you. What how do you define causality? The interesting thing about causality may be part of why maybe is a challenging thing to deal with particularly for statisticians I would say is that. It's very difficult to talk about it without finding yourself having a philosophical conversation and you know so going you know this is something that fill in. What is the causality? These in that philosophers have been wrestling with through the ages. Right hume had has counterfactual definition initial possibility. That's you know a follows from being had a not happy would not have happened But you know philosophers going back to the Buddha all kind of take their stab at what is caused -ality so there's a different philosophical arguments for causality and what it means I think from a practical standpoint. What most people mean when they say? causal inference is. They mean the estimation of Causal Effects. So if you're safer example at a tech company and you want to run some kind of experiment about the about whether a feature will drive a click or some other key performance indicator or metric. You're asking you. Your experiment is essentially trying to get at the question of what is the causal effect of this feature on this outcome and you'll be using the assumptions and methods from Statistics to estimate assuming Air Assumptions are valid those causal effects. But when we've talked in machine learning where now hearing you know. So I hadn Europe's like you said This talk about having agents that can understand that. Causal Structure of the world and and that causes allergies essential from moving from system one system to cognition day Pearl was very preeminent. Causal inference researcher talks about causal reasoning in in terms of free will and the ability to understand Dan intention and so there are definitely definitely a lot of angles to tackle this question from the perspective of artificial intelligence is that you know people who are running experiments in facebook. Netflix are not really thinking about

Coming up next