Well, I only got one of the two papers written that I was trying to write, so I've been asking myself exactly what a week is like. A week should be long enough to write two conference papers, especially since one already was in draft form by my coauthor (Philipp Rohlfshagen, who did a short postdoc with me and included a draft paper as one of his deliverables.)
I spent Saturday morning cleaning and then Will came Saturday afternoon so of course we went out and enjoyed ourselves Saturday. After having worked so late on Friday too (both of us) we spent Sunday at home mostly, just working (but also doing laundry.)
The conference I was submitting to was Cognitive Science, which this year had a special emphasis on the impact of ageing on cognition. I've had an interesting insight into ageing, learning and thought for a few years, since I was working on a task-learning problem, transitive inference (TI), with my friend Mark Baxter. But I couldn't work on it before may basic transitive paper got published (which happened a year ago now), and I haven't had time to work on it since then, because it doesn't have much to do with my fellowship topic. But I decided it was OK to dedicate one week to that and also to the Philipp paper and send them to the same conference. I originally had meant to do the Philipp paper first, then the TI paper if I had time. But I woke up Sunday with a great insight into how to write the TI paper and so started researching about that one. As I said, it was years since I'd had the idea so I needed to get up-to-date on the aging brain.
Then Monday I did a slightly-overdue review of a funding proposal for the BBSRC. I do review work in the order I get it, so I hadn't even looked at that before, but it turned out to be the best proposal I've ever read and I was really in a good mood for like two days after reading that, looking forward to seeing the science come out of it in a few years. Oh, and as last posted, I wound up going to an award ceremony with Will. We hadn't meant to, but I got email on short notice and it seemed the right thing to do. In fact, there really weren't enough people there given how big a deal the rewards were. So after an hour of incomprehensible talks and tributes and another hour of quite nice schmoozing over drinks (I met several professors with research interests near mine), we just went out and had a light dinner and then went to a cafe and worked a little.
Of course, I don't only do one thing a day, this is just a simple version -- there's a ton of email and management and negotiation you do every day -- I'm talking about editing a book with some people and I support the people back in Bath teaching my old courses (at least I hope they will give me new ones when I go back! I really, really, really want to teach something to do with my actual interests, AI. It's totally unbelievable Bath would hire an AI person from MIT and then never let her teach AI to their students.) Tuesday I met with my PhD student, Hagen, who had a family member die after a moderately long illness (it's been a source of stress for him for about 4 months) but decided to finish his (7 months overdue) 30-month viva this week before going to the funeral. We were mostly talking about the additional research he needs to do though. Like all British PhD students, he had only 3 years of funding which are now up, but he had a ton of distractions (also like most students) and is not finished (also like most students) and the question is when he can finish. He was a better-prepared academic than most UK PhD students coming in, so we both thought he would really finish in three years, but he also got even more disruptions (some of his own, some caused by outside things) so now it's a real problem for both of us changing our plans for the next year. He kept hoping I was wrong and he'd done enough work to finish quickly, but at the viva (which happened Thursday) he was convinced not. So we'll have to see what he can do in the next month and then replan...
Anyway, that was Tuesday morning. I met him in a cafe like usual, except that Will came too. Will also came to work with me on Monday, Wednesday and Friday, but Tuesday he stayed home and worked & also got laundry tokens from my building (this can only be done Tuesday afternoons!) The rest of Tuesday and Monday I spent researching aging brains.
Wednesday I paniced about the paper with Philipp and basically wound up writing it all in one day. I didn't quite manage to finish all the formatting. I had to make the contributions of the paper to cognitive science clear, and do all the referencing and formatting, and then get it short enough! I also had to understand exactly what he'd done, which was one of the reasons I needed to write the paper, to force myself to understand this, because I know I and I think possibly Hagen too will have to use what he did pretty soon. The paper is about solving an action-selection problem Hagen discovered with my system. It makes the decision making an AI animal makes more like real animals', by making them not stupidly persistent in a single goal. Older AI systems only dealt with one goal at a time, but that is not very useful in the real world (think of Asimov's three laws -- a robot that is going to get coffee still has to always make sure it is not allowing a human to be injured through it's inaction!) When I started my PhD newer AI systems tended to have more problem of switching between goals so quickly / reactively they didn't get much done. So I had introduced some persistence, but too much. Fixing that requires adding a tiny bit more cognition in -- recognizing when you've been interrupted and then reconsidering whether you should finish the goal you are currently working on. Actually, sounds like my week...
Thursday I had Hagen's viva in the morning, then attended a great talk at the university of Vienna, then went home to finish the paper, then met with Hagen, then went to a dinner party we had organized in a Vienna restaurant in honour of Will's birthday, mostly with people from the KLI but also one of my friends who does AI in Vienna came -- the first time I have seen her socially since I moved here! She gave us a chocolate cake. So Friday we brought that to the KLI, and Will got two pieces :-). But I never finished writing up my TI paper -- in fact, I barely started it! I was just too tired, which is I guess why I was so unhappy finishing the paper late the previous Friday -- I'm getting either too old or too unfit to work long Fridays now. I don't get as much exercise here as I did in Bath since I don't get to walk up that huge hill to work. Nor play squash or football! Anyway, so I missed the deadline.
Yesterday we slept until 11am, then we did a little shopping and then went out for a nice (private) Will birthday dinner. This morning we went to an art exhibit that was closing and then had a quick brunch at our favourite cafe, and now Will has gone to catch his plane.
Actually, I will be going to see him again soon, so I shouldn't be sad. Thursday and Friday he is teaching a course on inference for the social sciences which I will attend. But now I am torn as to whether to finish the TI paper. I had other plans for the coming week more in line with my fellowship, but it seems a waste to have figured all that material out and not write it down. But it is really not developed enough for a journal article. But surely there will be another conference I can submit it to...
I spent Saturday morning cleaning and then Will came Saturday afternoon so of course we went out and enjoyed ourselves Saturday. After having worked so late on Friday too (both of us) we spent Sunday at home mostly, just working (but also doing laundry.)
The conference I was submitting to was Cognitive Science, which this year had a special emphasis on the impact of ageing on cognition. I've had an interesting insight into ageing, learning and thought for a few years, since I was working on a task-learning problem, transitive inference (TI), with my friend Mark Baxter. But I couldn't work on it before may basic transitive paper got published (which happened a year ago now), and I haven't had time to work on it since then, because it doesn't have much to do with my fellowship topic. But I decided it was OK to dedicate one week to that and also to the Philipp paper and send them to the same conference. I originally had meant to do the Philipp paper first, then the TI paper if I had time. But I woke up Sunday with a great insight into how to write the TI paper and so started researching about that one. As I said, it was years since I'd had the idea so I needed to get up-to-date on the aging brain.
Then Monday I did a slightly-overdue review of a funding proposal for the BBSRC. I do review work in the order I get it, so I hadn't even looked at that before, but it turned out to be the best proposal I've ever read and I was really in a good mood for like two days after reading that, looking forward to seeing the science come out of it in a few years. Oh, and as last posted, I wound up going to an award ceremony with Will. We hadn't meant to, but I got email on short notice and it seemed the right thing to do. In fact, there really weren't enough people there given how big a deal the rewards were. So after an hour of incomprehensible talks and tributes and another hour of quite nice schmoozing over drinks (I met several professors with research interests near mine), we just went out and had a light dinner and then went to a cafe and worked a little.
Of course, I don't only do one thing a day, this is just a simple version -- there's a ton of email and management and negotiation you do every day -- I'm talking about editing a book with some people and I support the people back in Bath teaching my old courses (at least I hope they will give me new ones when I go back! I really, really, really want to teach something to do with my actual interests, AI. It's totally unbelievable Bath would hire an AI person from MIT and then never let her teach AI to their students.) Tuesday I met with my PhD student, Hagen, who had a family member die after a moderately long illness (it's been a source of stress for him for about 4 months) but decided to finish his (7 months overdue) 30-month viva this week before going to the funeral. We were mostly talking about the additional research he needs to do though. Like all British PhD students, he had only 3 years of funding which are now up, but he had a ton of distractions (also like most students) and is not finished (also like most students) and the question is when he can finish. He was a better-prepared academic than most UK PhD students coming in, so we both thought he would really finish in three years, but he also got even more disruptions (some of his own, some caused by outside things) so now it's a real problem for both of us changing our plans for the next year. He kept hoping I was wrong and he'd done enough work to finish quickly, but at the viva (which happened Thursday) he was convinced not. So we'll have to see what he can do in the next month and then replan...
Anyway, that was Tuesday morning. I met him in a cafe like usual, except that Will came too. Will also came to work with me on Monday, Wednesday and Friday, but Tuesday he stayed home and worked & also got laundry tokens from my building (this can only be done Tuesday afternoons!) The rest of Tuesday and Monday I spent researching aging brains.
Wednesday I paniced about the paper with Philipp and basically wound up writing it all in one day. I didn't quite manage to finish all the formatting. I had to make the contributions of the paper to cognitive science clear, and do all the referencing and formatting, and then get it short enough! I also had to understand exactly what he'd done, which was one of the reasons I needed to write the paper, to force myself to understand this, because I know I and I think possibly Hagen too will have to use what he did pretty soon. The paper is about solving an action-selection problem Hagen discovered with my system. It makes the decision making an AI animal makes more like real animals', by making them not stupidly persistent in a single goal. Older AI systems only dealt with one goal at a time, but that is not very useful in the real world (think of Asimov's three laws -- a robot that is going to get coffee still has to always make sure it is not allowing a human to be injured through it's inaction!) When I started my PhD newer AI systems tended to have more problem of switching between goals so quickly / reactively they didn't get much done. So I had introduced some persistence, but too much. Fixing that requires adding a tiny bit more cognition in -- recognizing when you've been interrupted and then reconsidering whether you should finish the goal you are currently working on. Actually, sounds like my week...
Thursday I had Hagen's viva in the morning, then attended a great talk at the university of Vienna, then went home to finish the paper, then met with Hagen, then went to a dinner party we had organized in a Vienna restaurant in honour of Will's birthday, mostly with people from the KLI but also one of my friends who does AI in Vienna came -- the first time I have seen her socially since I moved here! She gave us a chocolate cake. So Friday we brought that to the KLI, and Will got two pieces :-). But I never finished writing up my TI paper -- in fact, I barely started it! I was just too tired, which is I guess why I was so unhappy finishing the paper late the previous Friday -- I'm getting either too old or too unfit to work long Fridays now. I don't get as much exercise here as I did in Bath since I don't get to walk up that huge hill to work. Nor play squash or football! Anyway, so I missed the deadline.
Yesterday we slept until 11am, then we did a little shopping and then went out for a nice (private) Will birthday dinner. This morning we went to an art exhibit that was closing and then had a quick brunch at our favourite cafe, and now Will has gone to catch his plane.
Actually, I will be going to see him again soon, so I shouldn't be sad. Thursday and Friday he is teaching a course on inference for the social sciences which I will attend. But now I am torn as to whether to finish the TI paper. I had other plans for the coming week more in line with my fellowship, but it seems a waste to have figured all that material out and not write it down. But it is really not developed enough for a journal article. But surely there will be another conference I can submit it to...
Comments