Sullen today
Apr. 9th, 2019 07:41 amYesterday I overslept a bit. I finished the raw edges of the pockets and
prepared to hand-fell the side seams. Alliance continues; finally with some
better behavior from Ivan. At work I wrote a better script for user data
backups, fixed Apache, and tinkered with the sentence alignment generator
program I’m working on for B. F presented at the lab meeting, facing some
persistent issues with audience confusion about the task. We eventually got
everything straightened out: they’re studying games as a method of
evaluating risk-seeking vs risk-averse behavior. The ultimate goal is to
have a system that classifies a particular gameplay record (/policy), but
before they can do that, they have to design a game that sufficiently
differentiates the two extremes. So, they’re using prospect theory to
generate synthetic risk-seeking, risk-averse, and rational (term-of-art)
players, and using a Markov model to learn the reward function that best
separates the three resultant policies. Cool stuff. There’s some question
of whether gameplay is a good indicator of workplace behavior, and whether
this method is really all that much more scalable than a questionnaire, but
it’s a fun problem all the same.
Home, dinner, folded towels, Flash. I’ve been struggling with some social
anxiety, and coping by nominating rafts of women for my ugrad’s
presidential search committee and otherwise hiding. P got home just in time
for bedtime snuggles.
prepared to hand-fell the side seams. Alliance continues; finally with some
better behavior from Ivan. At work I wrote a better script for user data
backups, fixed Apache, and tinkered with the sentence alignment generator
program I’m working on for B. F presented at the lab meeting, facing some
persistent issues with audience confusion about the task. We eventually got
everything straightened out: they’re studying games as a method of
evaluating risk-seeking vs risk-averse behavior. The ultimate goal is to
have a system that classifies a particular gameplay record (/policy), but
before they can do that, they have to design a game that sufficiently
differentiates the two extremes. So, they’re using prospect theory to
generate synthetic risk-seeking, risk-averse, and rational (term-of-art)
players, and using a Markov model to learn the reward function that best
separates the three resultant policies. Cool stuff. There’s some question
of whether gameplay is a good indicator of workplace behavior, and whether
this method is really all that much more scalable than a questionnaire, but
it’s a fun problem all the same.
Home, dinner, folded towels, Flash. I’ve been struggling with some social
anxiety, and coping by nominating rafts of women for my ugrad’s
presidential search committee and otherwise hiding. P got home just in time
for bedtime snuggles.