Interview with RAD student, Daniel Angelov

One of our students, Daniel Angelov, has spent the past few months at Xerox PARC as an intern. He has been doing interesting things associated with the COGLE project I mentioned earlier, part of the DARPA XAI programme.

The folks at PARC have put up this interview on their blog, in which Daniel talks about his work and time there:

http://blogs.parc.com/2017/10/meet-the-parc-intern-daniel-angelov-talks-robots-and-the-future/

Advertisements

BBC World Service Radio – Click

We were on BBC World Service Radio’s Click programme last night: http://www.bbc.co.uk/programmes/p002w6r2, in a special episode on Robots for Humans, as part of their What If? series.

Unlike some other staged demos we have done in the past, for this one we deployed our current research work directly – algorithm(s) for learning to adapt to the unknown behaviour profile of the human player, work from our AAMAS papers and an upcoming UAI submission. It was a pleasure to see that its performance did genuinely surprise our hosts, Gareth and Bill (a bit more so in a rehearsal when Gareth first tried playing against our robot).

In my next post, I’ll discuss the problem and its context a bit more substantially – getting back into this blog after a bit of a holiday season hiatus!

Probability, causality and interaction

I just finished listening to this interview with Judea Pearl, which I enjoyed very much:

http://www.stephenibaraki.com/audio/Judea_Pearl.mp3

I particularly like the way his discussion fluidly weaves in philosophical issues of causality, counterfactuals, etc. with relatively practical issues of robots interacting with people. I am particularly glad to see someone of his (theoretical) stature take the notion and possibilities of human-agent interactions so seriously!

Interview on Flexibility Envelope and IT Conversations

Recently, I gave an interview to Per Sjoborg who runs a robotics related blog, covering topics ranging from modular robotics to transfer learning – the  emphasis being on the issue of dealing with change and new task contexts, etc. Here are links to that conversation:

http://flexibilityenvelope.com/interview-with-prof-ramamoorthy-about-free-agents-and-transfer-learning-in-control-of-srcmr-in-the-tenth-episode-of-the-flexible-elements-podcast

http://itc.conversationsnetwork.org/shows/detail5199.html

ET Interview with Prof. C.R. Rao

My colleague Michael Herrmann kicked off our Information Geometry reading group last week, with a discussion on the Cramer-Rao bound. We’re taking an easy tour around various areas of statistics and geometry before jumping into Amari’s book and related material.

By way of introduction, he mentioned an interview article (The ET Interview: Professor C.R. Rao, Econometric Theory 19:331-400, 2003) that I have now had the chance to read fully and like a lot.

Although the full article is an excellent read, I found a couple of his answers particularly interesting and worth reproducing here.

On keeping an open mind…

Q: You have worked on a vast number of topics covering almost every area of statistics. Is there any topic that you wanted to cover but could not? Any particular advice to young researchers?
A: One never knows. New uses of statistics are discovered from time to time. There are wide-open fields for young research workers to sow the seeds and reap the harvest.

On the importance of being a decent person…

Q: You also maintained a very good relationship with Neyman in spite of being a direct student of R.A. Fisher, with whom he had a bitter relationship. What is your secret? Also can you elaborate on the mutual relationships among Fisher, Egon Pearson, and J. Neyman? You are now the only bridge to that vanished world. Many people will be keenly interested in your recollections and observations.

A: There were bitter debates between Fisher and Neyman on testing of hypotheses and analysis of variance tests in design of experiments. In retrospect, we can see that there are lacuna in the theories propounded both by Fisher and Neyman. Perhaps much was due to conflict of personalities and Fisher’s intolerance to criticism of his work. Neyman was by disposition a very kind man. Although I criticized his work, he used to invite me to his home whenever I visited Berkeley. Neyman was the first to notice my work on the bound to variance of an unbiased estimate and gave it the name Cramér–Rao inequality. When I was visiting the University of Illinois in 1953–54, Neyman invited me to spend the summer in Berkeley and teach a course. When my wife and I reached Berkeley by train, he personally came to receive us and treated us kindly. He was also thinking of giving me a job in Berkeley, but for some reason this did not happen.