Related
I have some data about programmer actions inside an ide. From this data I am trying to make a good algorithm to calculate a programmers efficiency.
If we consider
efficiency = useful energy out / energy in
I made this rough equation:
energy in = active time(run events x code editing time)
Basically its the the time where stuff is actually being done by the programmer multiplied by run events like
debugging,build etc x the time where the programmer is actually editing code.
useful energy out = energy in - (#unsuccessfulbuilds + abortedtestruns
+ debuggerusetime)
Useful energy out is basically energy in minus things that I consider to be inefficient.
Can anyone see how to improve this, particularly from a mathematical point of view. Maths isn't my strong point and am not sure if I should use some sort of weighting for the equations and how to do this correctly. Also, I'm thinking of how to make it that whats minused from energy in in the useful energy out equation cant end up as less than 0. Can anyone give a hand with these questions ?
Your "algorithm" is completely arbitrary, making judgement of value over things that are inocuous to whatever you called "efficient/inneficient", and will endup with a completely incoherent final value after being calculated. Compilation time? So the first compilation of a C++ plugin that takes 30+ minutes is good? Debugging time is both efficient and inefficient in your proposal.
A programmer that codes for 10 minutes and make 6 consecutive builds with close to no changes will have the same output as the guy that code for 60 minutes.
I suggest you look firts to what is a good use of a programmers time, how other programs contabilize programmers efficiency. Etc.
Just on a side note, to create a model of efficiency of work of a highly technical and creative field, you must understand quite well math, statistics and project management. Thats why good scrum masters are so sought after.
Anyway, what you propose is not an algorithm, but a scoring system, usually algorithms do make use of scoring systems to help their internal rules work out the best solution based on the scoring. The scoring is just a value, while the algorithm is a process to an end.
I was wondering if anyone knew of a source which provides 2D model search spaces to test a GA against. I believe i read a while ago that there are a bunch of standard search spaces which are typically used when evaluating these type of algorithms.
If not, is it just a case of randomly generating this data yourself each time?
Edit: View from above and from the side.
The search space is completely dependent on your problem. The idea of a genetic algorithm being that modify the "genome" of a population of individuals to create the next generation, measure the fitness of the new generation and modify the genomes again with some randomness thrown is to try to prevent getting stuck in local minima. The search space however is completely determined by what you have in your genome, which in turn in completely determined by what the problem is.
There might be standard search spaces (i.e. genomes) that have been found to work well for particular problems (I haven't heard of any) but usually the hardest part in using GAs is defining what you have in your genome and how it is allowed to mutate. The usefulness comes from the fact that you don't have to explicitly declare all the values for the different variables for the model, but you can find good values (not necessarily the best ones though) using a more or less blind search.
EXAMPLE
One example used quite heavily is the evolved radio antenna (Wikipedia). The aim is to find a configuration for a radio antenna such that the antenna itself is as small and lightweight as possible, with the restriction that is has to respond to certain frequencies and have low noise and so on.
So you would build your genome specifying
the number of wires to use
the number of bends in each wire
the angle of each bend
maybe the distance of each bend from the base
(something else, I don't know what)
run your GA, see what comes out the other end, analyse why it didn't work. GAs have a habit of producing results you didn't expect because of bugs in the simulation. Anyhow, you discover that maybe the genome has to encode the number of bends individually for each of the wires in the antenna, meaning that the antenna isn't going to be symmetric. So you put that in your genome and run the thing again. Simulating stuff that needs to work in the physical world is usually the most expensive because at some point you have to test the indivudal(s) in the real world.
There's a reasonable tutorial of genetic algorithms here with some useful examples about different encoding schemes for the genome.
One final point, when people say that GAs are simple and easy to implement, they mean that the framework around the GA (generating a new population, evaluating fitness etc.) is simple. What usually is not said is that setting up a GA for a real problem is very difficult and usually requires a lot of trial and error because coming up with an encoding scheme that works well is not simple for complex problems. The best way to start is to start simple and make things more complex as you go along. You can of course make another GA to come with the encoding for first GA :).
There are several standard benchmark problems out there.
BBOB (Black Box Optimization Benchmarks) -- have been used in recent years as part of a continuous optimization competition
DeJong functions -- pretty old, and really too easy for most practical purposes these days. Useful for debugging perhaps.
ZDT/DTLZ multiobjective functions -- multi-objective optimization problems, but you could scalarize them yourself I suppose.
Many others
I did a little GP (note:very little) work in college and have been playing around with it recently. My question is in regards to the intial run settings (population size, number of generations, min/max depth of trees, min/max depth of initial trees, percentages to use for different reproduction operations, etc.). What is the normal practice for setting these parameters? What papers/sites do people use as a good guide?
You'll find that this depends very much on your problem domain - in particular the nature of the fitness function, your implementation DSL etc.
Some personal experience:
Large population sizes seem to work
better when you have a noisy fitness
function, I think this is because the growth
of sub-groups in the population over successive generations acts
to give more sampling of
the fitness function. I typically use
100 for less noisy/deterministic functions, 1000+
for noisy.
For number of generations it is best to measure improvements in the
fitness function and stop when it
meets your target criteria. I normally run a few hundred generations and see what kind of answers are coming out, if it is showing no improvement then you probably have an issue elsewhere.
Tree depth requirements are really dependent on your DSL. I sometimes try to do an
implementation without explicit
limits but penalise or eliminate
programs that run too long (which is probably
what you really care about....). I've also found total node counts of ~1000 to be quite useful hard limits.
Percentages for different mutation / recombination operators don't seem
to matter all that much. As long as
you have a comprehensive set of mutations, any reasonably balanced
distribution will usually work. I think the reason for this is that you are basically doing a search for favourable improvements so the main objective is just to make sure the trial improvements are reasonably well distributed across all the possibilities.
Why don't you try using a genetic algorithm to optimise these parameters for you? :)
Any problem in computer science can be
solved with another layer of
indirection (except for too many
layers of indirection.)
-David J. Wheeler
When I started looking into Genetic Algorithms I had the same question.
I wanted to collect data variating parameters on a very simple problem and link given operators and parameters values (such as mutation rates, etc) to given results in function of population size etc.
Once I started getting into GA a bit more I then realized that given the enormous number of variables this is a huge task, and generalization is extremely difficult.
talking from my (limited) experience, if you decide to simplify the problem and use a fixed way to implement crossover, selection, and just play with population size and mutation rate (implemented in a given way) trying to come up with general results you'll soon realize that too many variables are still into play because at the end of the day the number of generations after which statistically you will get a decent result (whatever way you wanna define decent) still obviously depend primarily on the problem you're solving and consequently on the genome size (representing the same problem in different ways will obviously lead to different results in terms of effect of given GA parameters!).
It is certainly possible to draft a set of guidelines - as the (rare but good) literature proves - but you will be able to generalize the results effectively in statistical terms only when the problem at hand can be encoded in the exact same way and the fitness is evaluated in a somehow an equivalent way (which more often than not means you're ealing with a very similar problem).
Take a look at Koza's voluminous tomes on these matters.
There are very different schools of thought even within the GP community -
Some regard populations in the (low) thousands as sufficient whereas Koza and others often don't deem if worthy to start a GP run with less than a million individuals in the GP population ;-)
As mentioned before it depends on your personal taste and experiences, resources and probably the GP system used!
Cheers,
Jan
I've been wondering if there are known solutions for algorithm of creating a school timetable. Basically, it's about optimizing "hour-dispersion" (both in teachers and classes case) for given class-subject-teacher associations. We can assume that we have sets of classes, lesson subjects and teachers associated with each other at the input and that timetable should fit between 8AM and 4PM.
I guess that there is probably no accurate algorithm for that, but maybe someone knows a good approximation or hints for developing it.
This problem is NP-Complete!
In a nutshell one needs to explore all possible combinations to find the list of acceptable solutions. Because of the variations in the circumstances in which the problem appears at various schools (for example: Are there constraints with regards to classrooms?, Are some of the classes split in sub-groups some of the time?, Is this a weekly schedule? etc.) there isn't a well known problem class which corresponds to all the scheduling problems. Maybe, the Knapsack problem has many elements of similarity with these problems at large.
A confirmation that this is both a hard problem and one for which people perennially seek a solution, is to check this (long) list of (mostly commercial) software scheduling tools
Because of the big number of variables involved, the biggest source of which are, typically, the faculty member's desires ;-)..., it is typically impractical to consider enumerating all possible combinations. Instead we need to choose an approach which visits a subset of the problem/solution spaces.
- Genetic Algorithms, cited in another answer is (or, IMHO, seems) well equipped to perform this kind of semi-guided search (The problem being to find a good evaluation function for the candidates to be kept for the next generation)
- Graph Rewriting approaches are also of use with this type of combinatorial optimization problems.
Rather than focusing on particular implementations of an automatic schedule generator program, I'd like to suggest a few strategies which can be applied, at the level of the definition of the problem.
The general rationale is that in most real world scheduling problems, some compromises will be required, not all constraints, expressed and implied: will be satisfied fully. Therefore we help ourselves by:
Defining and ranking all known constraints
Reducing the problem space, by manually, providing a set of additional constraints.This may seem counter-intuitive but for example by providing an initial, partially filled schedule (say roughly 30% of the time-slots), in a way that fully satisfies all constraints, and by considering this partial schedule immutable, we significantly reduce the time/space needed to produce candidate solutions. Another way additional constraints help is for example "artificially" adding a constraint which prevent teaching some subjects on some days of the week (if this is a weekly schedule...); this type of constraints results in reducing the problem/solution spaces, without, typically, excluding a significant number of good candidates.
Ensuring that some of the constraints of the problem can be quickly computed. This is often associated with the choice of data model used to represent the problem; the idea is to be able to quickly opt-for (or prune-out) some of the options.
Redefining the problem and allowing some of the constraints to be broken, a few times, (typically towards the end nodes of the graph). The idea here is to either remove some of constraints for filling-in the last few slots in the schedule, or to have the automatic schedule generator program stop shy of completing the whole schedule, instead providing us with a list of a dozen or so plausible candidates. A human is often in a better position to complete the puzzle, as indicated, possibly breaking a few of the contraints, using information which is not typically shared with the automated logic (eg "No mathematics in the afternoon" rule can be broken on occasion for the "advanced math and physics" class; or "It is better to break one of Mr Jones requirements than one of Ms Smith ... ;-) )
In proof-reading this answer , I realize it is quite shy of providing a definite response, but it none the less full of practical suggestions. I hope this help, with what is, after all, a "hard problem".
It's a mess. a royal mess. To add to the answers, already very complete, I want to point out my family experience. My mother was a teacher and used to be involved in the process.
Turns out that having a computer to do so is not only difficult to code per-se, it is also difficult because there are conditions that are difficult to specify to a pre-baked computer program. Examples:
a teacher teaches both at your school and at another institute. Clearly, if he ends the lesson there at 10.30, he cannot start at your premises at 10.30, because he needs some time to commute between the institutes.
two teachers are married. In general, it's considered good practice not to have two married teachers on the same class. These two teachers must therefore have two different classes
two teachers are married, and their child attends the same school. Again, you have to prevent the two teachers to teach in the specific class where their child is.
the school has separate facilities, like one day the class is in one institute, and another day the class is in another.
the school has shared laboratories, but these laboratories are available only on certain weekdays (for security reasons, for example, where additional personnel is required).
some teachers have preferences for the free day: some prefer on Monday, some on Friday, some on Wednesday. Some prefer to come early in the morning, some prefer to come later.
you should not have situations where you have a lesson of say, history at the first hour, then three hours of math, then another hour of history. It does not make sense for the students, nor for the teacher.
you should spread the arguments evenly. It does not make sense to have the first days in the week only math, and then the rest of the week only literature.
you should give some teachers two consecutive hours to do evaluation tests.
As you can see, the problem is not NP-complete, it's NP-insane.
So what they do is that they have a large table with small plastic insets, and they move the insets around until a satisfying result is obtained. They never start from scratch: they normally start from the previous year timetable and make adjustments.
The International Timetabling Competition 2007 had a lesson scheduling track and exam scheduling track. Many researchers participated in that competition. Lots of heuristics and metaheuristics were tried, but in the end the local search metaheuristics (such as Tabu Search and Simulated Annealing) clearly beat other algorithms (such as genetic algorithms).
Take a look at the 2 open source frameworks used by some of the finalists:
JBoss OptaPlanner (Java, open source)
Unitime (Java, open source) - more for universities
One of my half-term assignments was an genetic-algorithm school table generation.
Whole table is one "organism". There were some changes and caveats to the generic genetic algorithms approach:
Rules were made for "illegal tables": two classes in the same classroom, one teacher teaching two groups at the same time etc. These mutations were deemed lethal immediately and a new "organism" was sprouted in place of the "deceased" immediately. The initial one was generated by a series of random tries to get a legal (if senseless) one. Lethal mutation wasn't counted towards count of mutations in iteration.
"Exchange" mutations were much more common than "Modify" mutations. Changes were only between parts of the gene that made sense - no substituting a teacher with a classroom.
Small bonuses were assigned for bundling certain 2 hours together, for assigning same generic classroom in sequence for the same group, for keeping teacher's work hours and class' load continuous. Moderate bonuses were assigned for giving correct classrooms for given subject, keeping class hours within bonds (morning or afternoon), and such. Big bonuses were for assigning correct number of given subject, given workload for a teacher etc.
Teachers could create their workload schedules of "want to work then", "okay to work then", "doesn't like to work then", "can't work then", with proper weights assigned. Whole 24h were legal work hours except night time was very undesired.
The weight function... oh yeah. The weight function was huge, monstrous product (as in multiplication) of weights assigned to selected features and properties. It was extremely steep, one property easily able to change it by an order of magnitude up or down - and there were hundreds or thousands of properties in one organism. This resulted in absolutely HUGE numbers as the weights, and as a direct result, need to use a bignum library (gmp) to perform the calculations. For a small testcase of some 10 groups, 10 teachers and 10 classrooms, the initial set started with note of 10^-200something and finished with 10^+300something. It was totally inefficient when it was more flat. Also, the values grew a lot wider distance with bigger "schools".
Computation time wise, there was little difference between a small population (100) over a long time and a big population (10k+) over less generations. The computation over the same time produced about the same quality.
The calculation (on some 1GHz CPU) would take some 1h to stabilize near 10^+300, generating schedules that looked quite nice, for said 10x10x10 test case.
The problem is easily paralellizable by providing networking facility that would exchange best specimens between computers running the computation.
The resulting program never saw daylight outside getting me a good grade for the semester. It showed some promise but I never got enough motivation to add any GUI and make it usable to general public.
This problem is tougher than it seems.
As others have alluded to, this is a NP-complete problem, but let's analyse what that means.
Basically, it means you have to look at all possible combinations.
But "look at" doesn't tell you much what you need to do.
Generating all possible combinations is easy. It might produce a huge amount of data, but you shouldn't have much problems understanding the concepts of this part of the problem.
The second problem is the one of judging whether a given possible combination is good, bad, or better than the previous "good" solution.
For this you need more than just "is it a possible solution".
For instance, is the same teacher working 5 days a week for X weeks straight? Even if that is a working solution, it might not be a better solution than alternating between two people so that each teacher does one week each. Oh, you didn't think about that? Remember, this is people you're dealing with, not just a resource allocation problem.
Even if one teacher could work full-time for 16 weeks straight, that might be a sub-optimal solution compared to a solution where you try to alternate between teachers, and this kind of balancing is very hard to build into software.
To summarize, producing a good solution to this problem will be worth a lot, to many many people. Hence, it's not an easy problem to break down and solve. Be prepared to stake out some goals that aren't 100% and calling them "good enough".
My timetabling algorithm, implemented in FET (Free Timetabling Software, http://lalescu.ro/liviu/fet/ , a successful application):
The algorithm is heuristic. I named it "recursive swapping".
Input: a set of activities A_1...A_n and the constraints.
Output: a set of times TA_1...TA_n (the time slot of each activity. Rooms are excluded here, for simplicity). The algorithm must put each activity at a time slot, respecting constraints. Each TA_i is between 0 (T_1) and max_time_slots-1 (T_m).
Constraints:
C1) Basic: a list of pairs of activities which cannot be simultaneous (for instance, A_1 and A_2, because they have the same teacher or the same students);
C2) Lots of other constraints (excluded here, for simplicity).
The timetabling algorithm (which I named "recursive swapping"):
Sort activities, most difficult first. Not critical step, but speeds up the algorithm maybe 10 times or more.
Try to place each activity (A_i) in an allowed time slot, following the above order, one at a time. Search for an available slot (T_j) for A_i, in which this activity can be placed respecting the constraints. If more slots are available, choose a random one. If none is available, do recursive swapping:
a. For each time slot T_j, consider what happens if you put A_i into T_j. There will be a list of other activities which don't agree with this move (for instance, activity A_k is on the same slot T_j and has the same teacher or same students as A_i). Keep a list of conflicting activities for each time slot T_j.
b. Choose a slot (T_j) with lowest number of conflicting activities. Say the list of activities in this slot contains 3 activities: A_p, A_q, A_r.
c. Place A_i at T_j and make A_p, A_q, A_r unallocated.
d. Recursively try to place A_p, A_q, A_r (if the level of recursion is not too large, say 14, and if the total number of recursive calls counted since step 2) on A_i began is not too large, say 2*n), as in step 2).
e. If successfully placed A_p, A_q, A_r, return with success, otherwise try other time slots (go to step 2 b) and choose the next best time slot).
f. If all (or a reasonable number of) time slots were tried unsuccessfully, return without success.
g. If we are at level 0, and we had no success in placing A_i, place it like in steps 2 b) and 2 c), but without recursion. We have now 3 - 1 = 2 more activities to place. Go to step 2) (some methods to avoid cycling are used here).
UPDATE: from comments ... should have heuristics too!
I'd go with Prolog ... then use Ruby or Perl or something to cleanup your solution into a prettier form.
teaches(Jill,math).
teaches(Joe,history).
involves(MA101,math).
involves(SS104,history).
myHeuristic(D,A,B) :- [test_case]->D='<';D='>'.
createSchedule :- findall(Class,involves(Class,Subject),Classes),
predsort(myHeuristic,Classes,ClassesNew),
createSchedule(ClassesNew,[]).
createSchedule(Classes,Scheduled) :- [the actual recursive algorithm].
I am (still) in the process of doing something similar to this problem but using the same path as I just mentioned. Prolog (as a functional language) really makes solving NP-Hard problems easier.
Genetic algorithms are often used for such scheduling.
Found this example (Making Class Schedule Using Genetic Algorithm) which matches your requirement pretty well.
Here are a few links I found:
School timetable - Lists some problems involved
A Hybrid Genetic Algorithm for School Timetabling
Scheduling Utilities and Tools
This paper describes the school timetable problem and their approach to the algorithm pretty well: "The Development of SYLLABUS—An Interactive, Constraint-Based Scheduler for Schools and Colleges."[PDF]
The author informs me the SYLLABUS software is still being used/developed here: http://www.scientia.com/uk/
I work on a widely-used scheduling engine which does exactly this. Yes, it is NP-Complete; the best approaches seek to approximate an optimal solution. And, of course there are a lot of different ways to say which one is the "best" solution - is it more important that your teachers are happy with their schedules, or that students get into all their classes, for instance?
The absolute most important question you need to resolve early on is what makes one way of scheduling this system better than another? That is, if I have a schedule with Mrs Jones teaching Math at 8 and Mr Smith teaching Math at 9, is that better or worse than one with both of them teaching Math at 10? Is it better or worse than one with Mrs Jones teaching at 8 and Mr Jones teaching at 2? Why?
The main advice I'd give here is to divide the problem up as much as possible - maybe course by course, maybe teacher by teacher, maybe room by room - and work on solving the sub-problem first. There you should end up with multiple solutions to choose from, and need to pick one as the most likely optimal. Then, work on making the "earlier" sub-problems take into account the needs of later sub-problems in scoring their potential solutions. Then, maybe work on how to get yourself out of painted-into-the-corner situations (assuming you can't anticipate those situations in earlier sub-problems) when you get to a "no valid solutions" state.
A local-search optimization pass is often used to "polish" the end answer for better results.
Note that typically we are dealing with highly resource-constrained systems in school scheduling. Schools don't go through the year with a lot of empty rooms or teachers sitting in the lounge 75% of the day. Approaches which work best in solution-rich environments aren't necessarily applicable in school scheduling.
Generally, constraint programming is a good approach to this type of scheduling problem. A search on "constraint programming" and scheduling or "constraint based scheduling" both within stack overflow and on Google will generate some good references. It's not impossible - it's just a little hard to think about when using traditional optimization methods like linear or integer optimization. One output would be - does a schedule exist that satisfies all the requirements? That, in itself, is obviously helpful.
Good luck !
I have designed commercial algorithms for both class timetabling and examination timetabling. For the first I used integer programming; for the second a heuristic based on maximizing an objective function by choosing slot swaps, very similar to the original manual process that had been evolved. They main things in getting such solutions accepted are the ability to represent all the real-world constraints; and for human timetablers to not be able to see ways to improve the solution. In the end the algorithmic part was quite straightforward and easy to implement compared with the preparation of the databases, the user interface, ability to report on statistics like room utilization, user education and so on.
You can takle it with genetic algorithms, yes. But you shouldn't :). It can be too slow and parameter tuning can be too timeconsuming etc.
There are successful other approaches. All implemented in open source projects:
Constraint based approach
Implemented in UniTime (not really for schools)
You could also go further and use Integer programming. Successfully done at Udine university and also at University Bayreuth (I was involved there) using the commercial software (ILOG CPLEX)
Rule based approach with heuristisc - See Drools planner
Different heuristics - FET and my own
See here for a timetabling software list
I think you should use genetic algorithm because:
It is best suited for large problem instances.
It yields reduced time complexity on the price of inaccurate answer(Not the ultimate best)
You can specify constraints & preferences easily by adjusting fitness punishments for not met ones.
You can specify time limit for program execution.
The quality of solution depends on how much time you intend to spend solving the program..
Genetic Algorithms Definition
Genetic Algorithms Tutorial
Class scheduling project with GA
Also take a look at :a similar question and another one
This problem is MASSIVE where I work - imagine 1800 subjects/modules, and 350 000 students, each doing 5 to 10 modules, and you want to build an exam in 10 weeks, where papers are 1 hour to 3 days long... one plus point - all exams are online, but bad again, cannot exceed the system's load of max 5k concurrent. So yes we are doing this now in cloud on scaling servers.
The "solution" we used was simply to order modules on how many other modules they "clash" with descending (where a student does both), and to "backpack" them, allowing for these long papers to actually overlap, else it simply cannot be done.
So when things get too large, I found this "heuristic" to be practical... at least.
I don't know any one will agree with this code but i developed this code with the help of my own algorithm and is working for me in ruby.Hope it will help them who are searching for it
in the following code the periodflag ,dayflag subjectflag and the teacherflag are the hash with the corresponding id and the flag value which is Boolean.
Any issue contact me.......(-_-)
periodflag.each do |k2,v2|
if(TimetableDefinition.find(k2).period.to_i != 0)
subjectflag.each do |k3,v3|
if (v3 == 0)
if(getflag_period(periodflag,k2))
#teachers=EmployeesSubject.where(subject_name: #subjects.find(k3).name, division_id: division.id).pluck(:employee_id)
#teacherlists=Employee.find(#teachers)
teacherflag=Hash[teacher_flag(#teacherlists,teacherflag,flag).to_a.shuffle]
teacherflag.each do |k4,v4|
if(v4 == 0)
if(getflag_subject(subjectflag,k3))
subjectperiod=TimetableAssign.where("timetable_definition_id = ? AND subject_id = ?",k2,k3)
if subjectperiod.blank?
issubjectpresent=TimetableAssign.where("section_id = ? AND subject_id = ?",section.id,k3)
if issubjectpresent.blank?
isteacherpresent=TimetableAssign.where("section_id = ? AND employee_id = ?",section.id,k4)
if isteacherpresent.blank?
#finaltt=TimetableAssign.new
#finaltt.timetable_struct_id=#timetable_struct.id
#finaltt.employee_id=k4
#finaltt.section_id=section.id
#finaltt.standard_id=standard.id
#finaltt.division_id=division.id
#finaltt.subject_id=k3
#finaltt.timetable_definition_id=k2
#finaltt.timetable_day_id=k1
set_school_id(#finaltt,current_user)
if(#finaltt.save)
setflag_sub(subjectflag,k3,1)
setflag_period(periodflag,k2,1)
setflag_teacher(teacherflag,k4,1)
end
end
else
#subjectdetail=TimetableAssign.find_by_section_id_and_subject_id(#section.id,k3)
#finaltt=TimetableAssign.new
#finaltt.timetable_struct_id=#subjectdetail.timetable_struct_id
#finaltt.employee_id=#subjectdetail.employee_id
#finaltt.section_id=section.id
#finaltt.standard_id=standard.id
#finaltt.division_id=division.id
#finaltt.subject_id=#subjectdetail.subject_id
#finaltt.timetable_definition_id=k2
#finaltt.timetable_day_id=k1
set_school_id(#finaltt,current_user)
if(#finaltt.save)
setflag_sub(subjectflag,k3,1)
setflag_period(periodflag,k2,1)
setflag_teacher(teacherflag,k4,1)
end
end
end
end
end
end
end
end
end
end
end
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 11 years ago.
Genetic algorithms (GA) and genetic programming (GP) are interesting areas of research.
I'd like to know about specific problems you have solved using GA/GP and what libraries/frameworks you used if you didn't roll your own.
Questions:
What problems have you used GA/GP to solve?
What libraries/frameworks did you use?
I'm looking for first-hand experiences, so please do not answer unless you have that.
Not homework.
My first job as a professional programmer (1995) was writing a genetic-algorithm based automated trading system for S&P500 futures. The application was written in Visual Basic 3 [!] and I have no idea how I did anything back then, since VB3 didn't even have classes.
The application started with a population of randomly-generated fixed-length strings (the "gene" part), each of which corresponded to a specific shape in the minute-by-minute price data of the S&P500 futures, as well as a specific order (buy or sell) and stop-loss and stop-profit amounts. Each string (or "gene") had its profit performance evaluated by a run through 3 years of historical data; whenever the specified "shape" matched the historical data, I assumed the corresponding buy or sell order and evaluated the trade's result. I added the caveat that each gene started with a fixed amount of money and could thus potentially go broke and be removed from the gene pool entirely.
After each evaluation of a population, the survivors were cross-bred randomly (by just mixing bits from two parents), with the likelihood of a gene being selected as a parent being proportional to the profit it produced. I also added the possibility of point mutations to spice things up a bit. After a few hundred generations of this, I ended up with a population of genes that could turn $5000 into an average of about $10000 with no chance of death/brokeness (on the historical data, of course).
Unfortunately, I never got the chance to use this system live, since my boss lost close to $100,000 in less than 3 months trading the traditional way, and he lost his willingness to continue with the project. In retrospect, I think the system would have made huge profits - not because I was necessarily doing anything right, but because the population of genes that I produced happened to be biased towards buy orders (as opposed to sell orders) by about a 5:1 ratio. And as we know with our 20/20 hindsight, the market went up a bit after 1995.
I made a little critters that lived in this little world. They had a neural network brain which received some inputs from the world and the output was a vector for movement among other actions. Their brains were the "genes".
The program started with a random population of critters with random brains. The inputs and output neurons were static but what was in between was not.
The environment contained food and dangers. Food increased energy and when you have enough energy, you can mate. The dangers would reduce energy and if energy was 0, they died.
Eventually the creatures evolved to move around the world and find food and avoid the dangers.
I then decided to do a little experiment. I gave the creature brains an output neuron called "mouth" and an input neuron called "ear". Started over and was surprised to find that they evolved to maximize the space and each respective creature would stay in its respective part (food was placed randomly). They learned to cooperate with each other and not get in each others way. There were always the exceptions.
Then i tried something interesting. I dead creatures would become food. Try to guess what happened! Two types of creatures evolved, ones that attacked like in swarms, and ones that were high avoidance.
So what is the lesson here? Communication means cooperation. As soon as you introduce an element where hurting another means you gain something, then cooperation is destroyed.
I wonder how this reflects on the system of free markets and capitalism. I mean, if businesses can hurt their competition and get away with it, then its clear they will do everything in their power to hurt the competition.
Edit:
I wrote it in C++ using no frameworks. Wrote my own neural net and GA code. Eric, thank you for saying it is plausible. People usually don't believe in the powers of GA (although the limitations are obvious) until they played with it. GA is simple but not simplistic.
For the doubters, neural nets have been proven to be able to simulate any function if they have more than one layer. GA is a pretty simple way to navigate a solution space finding local and potentially global minimum. Combine GA with neural nets and you have a pretty good way to find functions that find approximate solutions for generic problems. Because we are using neural nets, then we are optimizing the function for some inputs, not some inputs to a function as others are using GA
Here is the demo code for the survival example: http://www.mempko.com/darcs/neural/demos/eaters/
Build instructions:
Install darcs, libboost, liballegro, gcc, cmake, make
darcs clone --lazy http://www.mempko.com/darcs/neural/
cd neural
cmake .
make
cd demos/eaters
./eaters
In January 2004, I was contacted by Philips New Display Technologies who were creating the electronics for the first ever commercial e-ink, the Sony Librie, who had only been released in Japan, years before Amazon Kindle and the others hit the market in US an Europe.
The Philips engineers had a major problem. A few months before the product was supposed to hit the market, they were still getting ghosting on the screen when changing pages. The problem was the 200 drivers that were creating the electrostatic field. Each of these drivers had a certain voltage that had to be set right between zero and 1000 mV or something like this. But if you changed one of them, it would change everything.
So optimizing each driver's voltage individually was out of the question. The number of possible combination of values was in billions,and it took about 1 minute for a special camera to evaluate a single combination. The engineers had tried many standard optimization techniques, but nothing would come close.
The head engineer contacted me because I had previously released a Genetic Programming library to the open-source community. He asked if GP/GA's would help and if I could get involved. I did, and for about a month we worked together, me writing and tuning the GA library, on synthetic data, and him integrating it into their system. Then, one weekend they let it run live with the real thing.
The following Monday I got these glowing emails from him and their hardware designer, about how nobody could believe the amazing results the GA found. This was it. Later that year the product hit the market.
I didn't get paid one cent for it, but I got 'bragging' rights. They said from the beginning they were already over budget, so I knew what the deal was before I started working on it. And it's a great story for applications of GAs. :)
I used a GA to optimize seating assignments at my wedding reception. 80 guests over 10 tables. Evaluation function was based on keeping people with their dates, putting people with something in common together, and keeping people with extreme opposite views at separate tables.
I ran it several times. Each time, I got nine good tables, and one with all the odd balls. In the end, my wife did the seating assignments.
My traveling salesman optimizer used a novel mapping of chromosome to itinerary, which made it trivial to breed and mutate the chromosomes without any risk of generating invalid tours.
Update: Because a couple people have asked how ...
Start with an array of guests (or cities) in some arbitrary but consistent ordering, e.g., alphabetized. Call this the reference solution. Think of a guest's index as his/her seat number.
Instead of trying to encode this ordering directly in the chromosome, we encode instructions for transforming the reference solution into a new solution. Specifically, we treat the chromosomes as lists of indexes in the array to swap. To get decode a chromosome, we start with the reference solution and apply all the swaps indicated by the chromosome. Swapping two entries in the array always results in a valid solution: every guest (or city) still appears exactly once.
Thus chromosomes can be randomly generated, mutated, and crossed with others and will always produce a valid solution.
I used genetic algorithms (as well as some related techniques) to determine the best settings for a risk management system that tried to keep gold farmers from using stolen credit cards to pay for MMOs. The system would take in several thousand transactions with "known" values (fraud or not) and figure out what the best combination of settings was to properly identify the fraudulent transactions without having too many false positives.
We had data on several dozen (boolean) characteristics of a transaction, each of which was given a value and totalled up. If the total was higher than a threshold, the transaction was fraud. The GA would create a large number of random sets of values, evaluate them against a corpus of known data, select the ones that scored the best (on both fraud detection and limiting the number of false positives), then cross breed the best few from each generation to produce a new generation of candidates. After a certain number of generations the best scoring set of values was deemed the winner.
Creating the corpus of known data to test against was the Achilles' heel of the system. If you waited for chargebacks, you were several months behind when trying to respond to the fraudsters, so someone would have to manually review large numbers of transactions to build up that corpus of data without having to wait too long.
This ended up identifying the vast majority of the fraud that came in, but couldn't quite get it below 1% on the most fraud-prone items (given that 90% of incoming transactions could be fraud, that was doing pretty well).
I did all this using perl. One run of the software on a fairly old linux box would take 1-2 hours to run (20 minutes to load data over a WAN link, the rest of the time spent crunching). The size of any given generation was limited by available RAM. I'd run it over and over with slight changes to the parameters, looking for an especially good result set.
All in all it avoided some of the gaffes that came with manually trying to tweak the relative values of dozens of fraud indicators, and consistently came up with better solutions than I could create by hand. AFAIK, it's still in use (about 3 years after I wrote it).
Football Tipping. I built a GA system to predict the week to week outcome of games in the AFL (Aussie Rules Football).
A few years ago I got bored of the standard work football pool, everybody was just going online and taking the picks from some pundit in the press. So, I figured it couldn't be too hard to beat a bunch of broadcast journalism majors, right? My first thought was to take the results from Massey Ratings and then reveal at the end of the season my strategy after winning fame and glory. However, for reasons I've never discovered Massey does not track AFL. The cynic in me believes it is because the outcome of each AFL game has basically become random chance, but my complaints of recent rule changes belong in a different forum.
The system basically considered offensive strength, defensive strength, home field advantage, week to week improvement (or lack thereof) and velocity of changes to each of these. This created a set of polynomial equations for each team over the season. The winner and score for each match for a given date could be computed. The goal was to find the set of coefficients that most closely matched the outcome of all past games and use that set to predict the upcoming weeks game.
In practice, the system would find solutions that accurately predicted over 90% of past game outcomes. It would then successfully pick about 60-80% of games for the upcoming week (that is the week not in the training set).
The result: just above middle of the pack. No major cash prize nor a system that I could use to beat Vegas. It was fun though.
I built everything from scratch, no framework used.
As well as some of the common problems, like the Travelling Salesman and a variation on Roger Alsing's Mona Lisa program, I've also written an evolutionary Sudoku solver (which required a bit more original thought on my part, rather than just re-implementing somebody else's idea). There are more reliable algorithms for solving Sudokus but the evolutionary approach works fairly well.
In the last few days I've been playing around with an evolutionary program to find "cold decks" for poker after seeing this article on Reddit. It's not quite satisfactory at the moment but I think I can improve it.
I have my own framework that I use for evolutionary algorithms.
I developed a home brew GA for a 3D laser surface profile system my company developed for the freight industry back in 1992.
The system relied upon 3 dimensional triangulation and used a custom laser line scanner, a 512x512 camera (with custom capture hw). The distance between the camera and laser was never going to be precise and the focal point of the cameras were not to be found in the 256,256 position that you expected it to be!
It was a nightmare to try and work out the calibration parameters using standard geometry and simulated annealing style equation solving.
The Genetic algorithm was whipped up in an evening and I created a calibration cube to test it on. I knew the cube dimensions to high accuracy and thus the idea was that my GA could evolve a set of custom triangulation parameters for each scanning unit that would overcome production variations.
The trick worked a treat. I was flabbergasted to say the least! Within around 10 generations my 'virtual' cube (generated from the raw scan and recreated from the calibration parameters) actually looked like a cube! After around 50 generations I had the calibration I needed.
Its often difficult to get an exact color combination when you are planning to paint your house. Often, you have some color in mind, but it is not one of the colors, the vendor shows you.
Yesterday, my Prof. who is a GA researcher mentioned about a true story in Germany (sorry, I have no further references, yes, I can find it out if any one requests to). This guy (let's call him the color guy) used to go from door-door to help people to find the exact color code (in RGB) that would be the closet to what the customer had in mind. Here is how he would do it:
The color guy used to carry with him a software program which used GA. He used to start with 4 different colors- each coded as a coded Chromosome (whose decoded value would be a RGB value). The consumer picks 1 of the 4 colors (Which is the closest to which he/she has in mind). The program would then assign the maximum fitness to that individual and move onto the next generation using mutation/crossover. The above steps would be repeated till the consumer had found the exact color and then color guy used to tell him the RGB combination!
By assigning maximum fitness to the color closes to what the consumer have in mind, the color guy's program is increasing the chances to converge to the color, the consumer has in mind exactly. I found it pretty fun!
Now that I have got a -1, if you are planning for more -1's, pls. elucidate the reason for doing so!
A couple of weeks ago, I suggested a solution on SO using genetic algorithms to solve a problem of graph layout. It is an example of a constrained optimization problem.
Also in the area of machine learning, I implemented a GA-based classification rules framework in c/c++ from scratch.
I've also used GA in a sample project for training artificial neural networks (ANN) as opposed to using the famous backpropagation algorithm.
In addition, and as part of my graduate research, I've used GA in training Hidden Markov Models as an additional approach to the EM-based Baum-Welch algorithm (in c/c++ again).
As part of my undergraduate CompSci degree, we were assigned the problem of finding optimal jvm flags for the Jikes research virtual machine. This was evaluated using the Dicappo benchmark suite which returns a time to the console. I wrote a distributed gentic alogirthm that switched these flags to improve the runtime of the benchmark suite, although it took days to run to compensate for hardware jitter affecting the results. The only problem was I didn't properly learn about the compiler theory (which was the intent of the assignment).
I could have seeded the initial population with the exisiting default flags, but what was interesting was that the algorithm found a very similar configuration to the O3 optimisation level (but was actually faster in many tests).
Edit: Also I wrote my own genetic algorithm framework in Python for the assignment, and just used the popen commands to run the various benchmarks, although if it wasn't an assessed assignment I would have looked at pyEvolve.
First off, "Genetic Programming" by Jonathan Koza (on amazon) is pretty much THE book on genetic and evolutionary algorithm/programming techniques, with many examples. I highly suggest checking it out.
As for my own use of a genetic algorithm, I used a (home grown) genetic algorithm to evolve a swarm algorithm for an object collection/destruction scenario (practical purpose could have been clearing a minefield). Here is a link to the paper. The most interesting part of what I did was the multi-staged fitness function, which was a necessity since the simple fitness functions did not provide enough information for the genetic algorithm to sufficiently differentiate between members of the population.
I am part of a team investigating the use of Evolutionary Computation (EC) to automatically fix bugs in existing programs. We have successfully repaired a number of real bugs in real world software projects (see this project's homepage).
We have two applications of this EC repair technique.
The first (code and reproduction information available through the project page) evolves the abstract syntax trees parsed from existing C programs and is implemented in Ocaml using our own custom EC engine.
The second (code and reproduction information available through the project page), my personal contribution to the project, evolves the x86 assembly or Java byte code compiled from programs written in a number of programming languages. This application is implemented in Clojure and also uses its own custom built EC engine.
One nice aspect of Evolutionary Computation is the simplicity of the technique makes it possible to write your own custom implementations without too much difficulty. For a good freely available introductory text on Genetic Programming see the Field Guide to Genetic Programming.
A coworker and I are working on a solution for loading freight onto trucks using the various criteria our company requires. I've been working on a Genetic Algorithm solution while he is using a Branch And Bound with aggressive pruning. We are still in the process of implementing this solution but so far, we have been getting good results.
Several years ago I used ga's to optimize asr (automatic speech recognition) grammars for better recognition rates. I started with fairly simple lists of choices (where the ga was testing combinations of possible terms for each slot) and worked my way up to more open and complex grammars. Fitness was determined by measuring separation between terms/sequences under a kind of phonetic distance function. I also experimented with making weakly equivalent variations on a grammar to find one that compiled to a more compact representation (in the end I went with a direct algorithm, and it drastically increased the size of the "language" that we could use in applications).
More recently I have used them as a default hypothesis against which to test the quality of solutions generated from various algorithms. This has largely involved categorization and different kinds of fitting problems (i.e. create a "rule" that explains a set of choices made by reviewers over a dataset(s)).
I made a complete GA framework named "GALAB", to solve many problems:
locating GSM ANTs (BTS) to decrease overlap & blank locations.
Resource constraint project scheduling.
Evolutionary picture creation. (Evopic)
Travelling salesman problem.
N-Queen & N-Color problems.
Knight's tour & Knapsack problems.
Magic square & Sudoku puzzles.
string compression, based on Superstring problem.
2D Packaging problem.
Tiny artificial life APP.
Rubik puzzle.
I once used a GA to optimize a hash function for memory addresses. The addresses were 4K or 8K page sizes, so they showed some predictability in the bit pattern of the address (least significant bits all zero; middle bits incrementing regularly, etc.) The original hash function was "chunky" - it tended to cluster hits on every third hash bucket. The improved algorithm had a nearly perfect distribution.
I built a simple GA for extracting useful patterns out of the frequency spectrum of music as it was being played. The output was used to drive graphical effects in a winamp plugin.
Input: a few FFT frames (imagine a 2D array of floats)
Output: single float value (weighted sum of inputs), thresholded to 0.0 or 1.0
Genes: input weights
Fitness function: combination of duty cycle, pulse width and BPM within sensible range.
I had a few GAs tuned to different parts of the spectrum as well as different BPM limits, so they didn't tend to converge towards the same pattern. The outputs from the top 4 from each population were sent to the rendering engine.
An interesting side effect was that the average fitness across the population was a good indicator for changes in the music, although it generally took 4-5 seconds to figure it out.
I don't know if homework counts...
During my studies we rolled our own program to solve the Traveling Salesman problem.
The idea was to make a comparison on several criteria (difficulty to map the problem, performance, etc) and we also used other techniques such as Simulated annealing.
It worked pretty well, but it took us a while to understand how to do the 'reproduction' phase correctly: modeling the problem at hand into something suitable for Genetic programming really struck me as the hardest part...
It was an interesting course since we also dabbled with neural networks and the like.
I'd like to know if anyone used this kind of programming in 'production' code.
I used a simple genetic algorithm to optimize the signal to noise ratio of a wave that was represented as a binary string. By flipping the the bits certain ways over several million generations I was able to produce a transform that resulted in a higher signal to noise ratio of that wave. The algorithm could have also been "Simulated Annealing" but was not used in this case. At their core, genetic algorithms are simple, and this was about as simple of a use case that I have seen, so I didn't use a framework for generation creation and selection - only a random seed and the Signal-to-Noise Ratio function at hand.
As part of my thesis I wrote a generic java framework for the multi-objective optimisation algorithm mPOEMS (Multiobjective prototype optimization with evolved improvement steps), which is a GA using evolutionary concepts. It is generic in a way that all problem-independent parts have been separated from the problem-dependent parts, and an interface is povided to use the framework with only adding the problem-dependent parts. Thus one who wants to use the algorithm does not have to begin from zero, and it facilitates work a lot.
You can find the code here.
The solutions which you can find with this algorithm have been compared in a scientific work with state-of-the-art algorithms SPEA-2 and NSGA, and it has been proven that
the algorithm performes comparable or even better, depending on the metrics you take to measure the performance, and especially depending on the optimization-problem you are looking on.
You can find it here.
Also as part of my thesis and proof of work I applied this framework to the project selection problem found in portfolio management. It is about selecting the projects which add the most value to the company, support most the strategy of the company or support any other arbitrary goal. E.g. selection of a certain number of projects from a specific category, or maximization of project synergies, ...
My thesis which applies this framework to the project selection problem:
http://www.ub.tuwien.ac.at/dipl/2008/AC05038968.pdf
After that I worked in a portfolio management department in one of the fortune 500, where they used a commercial software which also applied a GA to the project selection problem / portfolio optimization.
Further resources:
The documentation of the framework:
http://thomaskremmel.com/mpoems/mpoems_in_java_documentation.pdf
mPOEMS presentation paper:
http://portal.acm.org/citation.cfm?id=1792634.1792653
Actually with a bit of enthusiasm everybody could easily adapt the code of the generic framework to an arbitrary multi-objective optimisation problem.
At work I had the following problem: given M tasks and N DSPs, what was the best way to assign tasks to DSPs? "Best" was defined as "minimizing the load of the most loaded DSP". There were different types of tasks, and various task types had various performance ramifications depending on where they were assigned, so I encoded the set of job-to-DSP assignments as a "DNA string" and then used a genetic algorithm to "breed" the best assignment string I could.
It worked fairly well (much better than my previous method, which was to evaluate every possible combination... on non-trivial problem sizes, it would have taken years to complete!), the only problem was that there was no way to tell if the optimal solution had been reached or not. You could only decide if the current "best effort" was good enough, or let it run longer to see if it could do better.
There was an competition on codechef.com (great site by the way, monthly programming competitions) where one was supposed to solve an unsolveable sudoku (one should come as close as possible with as few wrong collumns/rows/etc as possible).What I would do, was to first generate a perfect sudoku and then override the fields, that have been given. From this pretty good basis on I used genetic programming to improve my solution.I couldn't think of a deterministic approach in this case, because the sudoku was 300x300 and search would've taken too long.
In a seminar in the school, we develop an application to generate music based in the musical mode. The program was build in Java and the output was a midi file with the song. We using distincts aproachs of GA to generate the music. I think this program can be useful to explore new compositions.
in undergrad, we used NERO (a combination of neural network and genetic algorithm) to teach in-game robots to make intelligent decisions. It was pretty cool.
I developed a multithreaded swing based simulation of robot navigation through a set of randomized grid terrain of food sources and mines and developed a genetic algorithm based strategy of exploring the optimization of robotic behavior and survival of fittest genes for a robotic chromosome. This was done using charting and mapping of each iteration cycle.
Since, then I have developed even more game behavior. An example application I built recently for myself was a genetic algorithm for solving the traveling sales man problem in route finding in UK taking into account start and goal states as well as one/multiple connection points, delays, cancellations, construction works, rush hour, public strikes, consideration between fastest vs cheapest routes. Then providing a balanced recommendation for the route to take on a given day.
Generally, my strategy is to use POJO based representaton of genes then I apply specific interface implementations for selection, mutation, crossover strategies, and the criteria point. My fitness function then basically becomes a quite complex based on the strategy and criteria I need to apply as a heuristic measure.
I have also looked into applying genetic algorithm into automated testing within code using systematic mutation cycles where the algorithm understands the logic and tries to ascertain a bug report with recommendations for code fixes. Basically, a way to optimize my code and provide recommendations for improvement as well as a way of automating the discovery of new programmatic code. I have also tried to apply genetic algorithms to music production amongst other applications.
Generally, I find evolutionary strategies like most metaheuristic/global optimization strategies, they are slow to learn at first but start to pick up as the solutions become closer and closer to goal state and as long as your fitness function and heuristics are well aligned to produce that convergence within your search space.
I once tried to make a computer player for the game of Go, exclusively based on genetic programming. Each program would be treated as an evaluation function for a sequence of moves. The programs produced weren't very good though, even on a rather diminuitive 3x4 board.
I used Perl, and coded everything myself. I would do things differently today.
After reading The Blind Watchmaker, I was interested in the pascal program Dawkins said he had developed to create models of organisms that could evolve over time. I was interested enough to write my own using Swarm. I didn't make all the fancy critter graphics he did, but my 'chromosomes' controlled traits which affected organisms ability to survive. They lived in a simple world and could slug it out against each other and their environment.
Organisms lived or died partly due to chance, but also based on how effectively they adapted to their local environments, how well they consumed nutrients & how successfully they reproduced. It was fun, but also more proof to my wife that I am a geek.
It was a while ago, but I rolled a GA to evolve what were in effect image processing kernels to remove cosmic ray traces from Hubble Space Telescope (HST) images. The standard approach is to take multiple exposures with the Hubble and keep only the stuff that is the same in all the images. Since HST time is so valuable, I'm an astronomy buff, and had recently attended the Congress on Evolutionary Computation, I thought about using a GA to clean up single exposures.
The individuals were in the form of trees that took a 3x3 pixel area as input, performed some calculations, and produced a decision about whether and how to modify the center pixel. Fitness was judged by comparing the output with an image cleaned up in the traditional way (i.e. stacking exposures).
It actually sort of worked, but not well enough to warrant foregoing the original approach. If I hadn't been time-constrained by my thesis, I might have expanded the genetic parts bin available to the algorithm. I'm pretty sure I could have improved it significantly.
Libraries used: If I recall correctly, IRAF and cfitsio for astronomical image data processing and I/O.
I experimented with GA in my youth. I wrote a simulator in Python that worked as follows.
The genes encoded the weights of a neural network.
The neural network's inputs were "antennae" that detected touches. Higher values meant very close and 0 meant not touching.
The outputs were to two "wheels". If both wheels went forward, the guy went forward. If the wheels were in opposite directions, the guy turned. The strength of the output determined the speed of the wheel turning.
A simple maze was generated. It was really simple--stupid even. There was the start at the bottom of the screen and a goal at the top, with four walls in between. Each wall had a space taken out randomly, so there was always a path.
I started random guys (I thought of them as bugs) at the start. As soon as one guy reached the goal, or a time limit was reached, the fitness was calculated. It was inversely proportional to the distance to the goal at that time.
I then paired them off and "bred" them to create the next generation. The probability of being chosen to be bred was proportional to its fitness. Sometimes this meant that one was bred with itself repeatedly if it had a very high relative fitness.
I thought they would develop a "left wall hugging" behavior, but they always seemed to follow something less optimal. In every experiment, the bugs converged to a spiral pattern. They would spiral outward until they touched a wall to the right. They'd follow that, then when they got to the gap, they'd spiral down (away from the gap) and around. They would make a 270 degree turn to the left, then usually enter the gap. This would get them through a majority of the walls, and often to the goal.
One feature I added was to put in a color vector into the genes to track relatedness between individuals. After a few generations, they'd all be the same color, which tell me I should have a better breeding strategy.
I tried to get them to develop a better strategy. I complicated the neural net--adding a memory and everything. It didn't help. I always saw the same strategy.
I tried various things like having separate gene pools that only recombined after 100 generations. But nothing would push them to a better strategy. Maybe it was impossible.
Another interesting thing is graphing the fitness over time. There were definite patterns, like the maximum fitness going down before it would go up. I have never seen an evolution book talk about that possibility.