{"id":153,"date":"2013-08-07T21:03:02","date_gmt":"2013-08-08T01:03:02","guid":{"rendered":"http:\/\/sites.hampshire.edu\/ci-lab\/?p=153"},"modified":"2013-08-07T21:03:02","modified_gmt":"2013-08-08T01:03:02","slug":"16-november-2012","status":"publish","type":"post","link":"https:\/\/sites.hampshire.edu\/ci-lab\/2013\/08\/07\/16-november-2012\/","title":{"rendered":"16 November 2012"},"content":{"rendered":"

Meeting Notes<\/h2>\n

next week we need to have a meeting, ONLINE!!!!<\/p>\n

Gecco deadline is January we should have results and bits of text by next week<\/p>\n

“Man will do anything in order to avoid the true difficulty of thinking” -proverb by Lee<\/p>\n

Go Zeke’s publication for Chemistry!<\/p>\n

Options with Ranjan<\/p>\n

Do linear regression guided by Ranjan on problems in chemistry that we have no idea about<\/p>\n

or we could try and evolve a material (organic or inorganic), which is evolved with fabrication and physical testing<\/p>\n

– takes about two days to do each generation<\/p>\n

– we need to focus on making each result good, rather then a hodgepodge of garbage with gems in it<\/p>\n

– Fitness prediction would be nice!<\/p>\n

– in between: simulate enzymes and run GP on them, no physical testing<\/p>\n

 <\/p>\n

Ranjan’s STUFFFFF<\/p>\n

-mechanistic deterministic molding of biological and chemical systems<\/p>\n

– has a teammate (Chris Corneleus) who can do polymers and stuff, but Ranjan doesn’t have supplies to do that<\/p>\n

– there’s a big initiative to do materials stuff called “materials genome”<\/p>\n

– Weird stuff happens when you mix polymers, it is not always the geometric mean, definitely non-intuitive and non-linear<\/p>\n

– It is mostly explored with intuition<\/p>\n

– applications range: water filtration, fuel stuff,<\/p>\n

– we want problems where “the proof is in the pudding”, so this is bawler<\/p>\n

– we would be making recipes for these polymers<\/p>\n

– the throughput is hella low throughput, so we need to make sure that it is unusually likely to land you in good places of a searchspace, and mutates\/crosses you over towards that<\/p>\n

– note we are evolving every part of the recipe<\/p>\n

– based on experience manipleting the temperature matters a lot<\/p>\n

– the starting components and processing are both incredibly important (particularly looking at water filtration)<\/p>\n

– the tests would be assessment of water purity, flow output (diffusion of the polymer), swelling of polymers<\/p>\n

– so non-linear that regression sucks like hell, the systems are multidimensional and non-linear<\/p>\n

– ok, not nessisairily the case that symbolic regression has not been tested with it<\/p>\n

– there are weird metrics like torchuosity, which is a metric of how strangely a molecule moves<\/p>\n

– potentially be useful to run grammatical evolution or developmental evolution (so the programs can be weird, by changing stuff and such things, but the starting seed is always good)<\/p>\n

– the seed formula is take two material mix them, heat them, cool them<\/p>\n

– getting the max yield from a small sample size is very, very, very important<\/p>\n

– check out evolutionary robotics field<\/p>\n

– simple story there is a shit ton of fitness cases, and then they look at situations in which the population varies significantly and then only looks at those fitness cases<\/p>\n

– unfortunately that is not a perfect match<\/p>\n

– josh bongaurd used a simulation for most of the tests, but would occasionally do a real life test which they would use to improve the simulation<\/p>\n

– we could try to evolve the simulator using fitness predictors<\/p>\n

-yay, there’s a lot of data already!<\/p>\n

– there is a lot of proprietary going into parrot front<\/p>\n

– there’s some data already<\/p>\n

-probs at least on the scale of a few hundred<\/p>\n

– that’s enough to at least to start<\/p>\n

– making something to predict efficacy of future recipes based on current recipes and their results<\/p>\n

– what step was x or y<\/p>\n

– try lots of heuristics<\/p>\n

– want to make sure we give the real test predicted solutions which are distinct, not just successful ones<\/p>\n

– do we keep individuals from the last test in the new test<\/p>\n

– Solubility is unpredictable, simulated spectra would be nice, we could do it, there’s also a shit ton of data on it already<\/p>\n

– we could do attempt something to do with enzymes and protein folding and doing predictions (sounds scary)<\/p>\n

– what if we could make something sufficiently good to do a denovo<\/p>\n

– in protiens there is a sequence space, a shape space and a function space, they’re hella weird to predict and are not always super successful attempts<\/p>\n

– they’re a bunch of stuff which suggests that shapespace may not even exist<\/p>\n

-would this be from data or simulation<\/p>\n

– it’s not got much data, but if you choose the right problem it could work<\/p>\n

 <\/p>\n

Lee’s writing on the board: zee’s writing – we start with data that goes like { recepe: measurement }, \u2026<\/p>\n

we have a population of R:M pairs<\/p>\n

we evolve a function that given an R gives an M<\/p>\n

we need to figure out how to chew up R<\/p>\n

that fitness predictor is used in another found of evolution<\/p>\n

where we evolve programs that change a recipe to maximize it’s fitness in measurements<\/p>\n

then we physically test a hundred of them, selected somehow<\/p>\n

now we have a new (larger) set of RM pairs
\n-a shit ton of ideas about how this should work<\/p>\n

– fitness predictors (how many, do we do lexicase selection, do we do multiple runs)<\/p>\n

– should we choose a diverse set of programs from the fitness predictor set (to clarify one’s that are distinct and will produce interesting darts on the search space)
\nwhat are scopes good for?<\/p>\n

gsxo? kata bowling?<\/p>\n

when you don’t have scopes it seems clearly like a bad idea to do something that operates not the whole stack.<\/p>\n

code map does crazy shit<\/p>\n

could also be used for strings<\/p>\n

Pre-Meeting Agenda<\/h2>\n