Algorithm to get probability of reaching a goal - algorithm

Okay, I'm gonna be as detailed as possible here.
Imagine the user goes through a set of 'options' he can choose. Every time he chooses, he get, say, 4 different options. There are many more options that can appear in those 4 'slots'. Each of those has a certain definite and known probability of appearing. Not all options are equally probable to appear, and some options require others to have already been selected previously - in a complex interdependence tree. (this I have already defined)
When the user chooses one of the 4, he is presented another choice of 4 options. The pool of options is defined again and can depend on what the user has chosen previously.
Among all possible 'options' that can ever appear, there are a certain select few which are special, call them KEY options.
When the program starts, the user is presented the first 4 options. For every one of those 4, the program needs to compute the total probability that the user will 'achieve' all the KEY options in a period of (variable) N choices.
e.g. if there are 4 options altogether the probability of achieving any one of them is exactly 1 since all of them appear right at the beginning.
If anyone can advise me as to what logic i should start with, I'd be very grateful.
I was thinking of counting all possible choice sequences, and counting the ones resulting in KEY options being chosen within N 'steps', but the problem is the probability is not uniform for all of them to appear, and also the pool of options changes as the user chooses and accumulates his options.
I'm having difficulty implementing the well defined probabilities and dependencies of the options into an algorithm that can give sensible total probability. So the user knows each time which of the 4 puts him in the best position to eventually acquire the KEY options.
Any ideas?
EDIT:
here's an example:
say there are 7 options in the pool. option1, ..., option7
option7 requires option6; option6 requires option4 and option5;
option1 thru 5 dont require anything and can appear immediately, with respective probabilities option1.p, ..., option5.p;
the KEY option is, say, option7;
user gets 4 randomly (but weighted) chosen options among 1-5, and the program needs to say something like:
"if you choose (first), you have ##% chance of getting option7 in at most N tries." analogous for the other 3 options.
naturally, for some low N it is impossible to get option7, and for some large N it is certain. N can be chosen but is fixed.
EDIT: So, the point here is NOT the user chooses randomly. Point is - the program suggests which option to choose, as to maximize the probability that eventually, after N steps, the user will be offered all key options.
For the above example; say we choose N = 4. so the program needs to tell us which of the first 4 options that appeared (any 4 among option1-5), which one, when chosen, yields the best chance of obtaining option7. since for option7 you need option6, and for that you need option4 and option5, it is clear that you MUST select either option4 or option5 on the first set of choices. one of them is certain to appear, of course.
Let's say we get this for the first choice {option3, option5, option2, option4}. The program then says:
if you chose option3, you'll never get option7 in 4 steps. p = 0;
if you chose option5, you might get option7, p=....;
... option2, p = 0;
... option4, p = ...;
Whatever we choose, for the next 4 options, the p's are re calculated. Clearly, if we chose option3 or option2, every further choice has exactly 0 probability of getting us to option7. But for option4 and option5, p > 0;
Is it clearer now? I don't know how to getting these probabilities p.

This sounds like a moderately fiddly Markov chain type problem. Create a node for every state; a state has no history, and is just dependent on the possible paths out of it (each weighted with some probability). You put a probability on each node, the chance that the user is in that state, so, for the first step, there will be a 1 his starting node, 0 everywhere else. Then, according to which nodes are adjacent and the chances of getting to them, you iterate to the next step by updating the probabilities on each vertex. So, you can calculate easily which states the user could land on in, say, 15 steps, and the associated probabilities. If you are interested in asymptotic behaviour (what would happen if he could play forever), you make a big pile of linear simultaneous equations and just solve them directly or using some tricks if your tree or graph has a neat form. You often end up with cyclical solutions, where the user could get stuck in a loop, and so on.

If you think the user selects the options at random, and he is always presented the same distribution of options at a node, you model this as a random walk on a graph. There was a recent nice post on calculating terminating probabilities of a particular random walks on the mathematica blog.

Related

Using Softmax Explorer (--cb_explore_adf) for Ranking in VowpalWabbit

I am trying to use VW to perform ranking using the contextual bandit framework, specifically using --cb_explore_adf --softmax --lambda X. The choice of softmax is because, according to VW's docs: "This is a different explorer, which uses the policy not only to predict an action but also predict a score indicating the quality of each action." This quality-related score is what I would like to use for ranking.
The scenario is this: I have a list of items [A, B, C, D], and I would like to sort it in an order that maximizes a pre-defined metric (e.g., CTR). One of the problems, as I see, is that we cannot evaluate the items individually because we can't know for sure which item made the user click or not.
To test some approaches, I've created a dummy dataset. As a way to try and solve the above problem, I am using the entire ordered list as a way to evaluate if a click happens or not (e.g., given the context for user X, he will click if the items are [C, A, B, D]). Then, I reward the items individually according to their position on the list, i.e., reward = 1/P for 0 < P < len(list). Here, the reward for C, A, B, D is 1, 0.5, and 0.25, 0.125, respectively. If there's no click, the reward is zero for all items. The reasoning behind this is that more important items will stabilize on top and less important on the bottom.
Also, one of the difficulties I found was defining a sampling function for this approach. Typically, we're interested in selecting only one option, but here I have to sample multiple times (4 in the example). Because of that, it's not very clear how I should incorporate exploration when sampling items. I have a few ideas:
Copy the probability mass function and assign it to copy_pmf. Draw a random number between 0 and max(copy_pmf) and for each probability value in copy_pmf, increment the sum_prob variable (very similar to the tutorial here:https://vowpalwabbit.org/tutorials/cb_simulation.html). When sum_prob > draw, we add the current item/prob to a list. Then, we remove this probability from copy_pmf, set sum_prob = 0, and draw a new number again between 0 and max(copy_pmf) (which might change or not).
Another option is drawing a random number and, if the maximum probability, i.e., max(pmf) is greater than this number, we exploit. If it isn't, we shuffle the list and return this (explore). This approach requires tuning the lambda parameter, which controls the output pmf (I have seen cases where the max prob is > 0.99, which would mean around a 1% chance of exploring. I have also seen instances where max prob is ~0.5, which is around 50% exploration.
I would like to know if there are any suggestions regarding this problem, specifically sampling and the reward function. Also, if there are any things I might be missing here.
Thank you!
That sounds like something that can be solved by conditional contextual bandits
For demo scenario that you are mentioning each example should have 4 slots.
You can use any exploration algorithm in this case and it is going to be done independently per each slot. Learning objective is average loss over all slots, but decisions are made sequentially from the first slot to the last, so you'll effectively learn the ranking even in case of binary reward here.

Jira's Lexorank algorithm for new stories

I am looking to create a large list of items that allows for easy insertion of new items and for easily changing the position of items within that list. When updating the position of an item, I want to change as few fields as possible regarding the order of items.
After some research, I found that Jira's Lexorank algorithm fulfills all of these needs. Each story in Jira has a 'rank-field' containing a string which is built up of 3 parts: <bucket>|<rank>:<sub-rank>. (I don't know whether these parts have actual names, this is what I will call them for ease of reference)
Examples of valid rank-fields:
0|vmis7l:hl4
0|i000w8:
0|003fhy:zzzzzzzzzzzw68bj
When dragging a card above 0|vmis7l:hl4, the new card will receive rank 0|vmis7l:hl2, which means that only the rank-field for this new card needs to be updated while the entire list can always be sorted on this rank-field. This is rather clever, and I can't imagine that Lexorank is the only algorithm to use this.
Is there a name for this method of sorting used in the sub-rank?
My question is related to the creation of new cards in Jira. Each new card starts with an empty sub-rank, and the rank is always chosen such that the new card is located at the bottom of the list. I've created a bunch of new stories just to see how the rank would change, and it seems that the rank is always incremented by 8 (in base-36).
Does anyone know more specifically how the rank for new cards is generated? Why is it incremented by 8?
I can only imagine that after some time (270 million cards) there are no more ranks to generate, and the system needs to recalculate the rank-field of all cards to make room for additional ranks.
Are there other triggers that require recalculation of all rank-fields?
I suppose the bucket plays a role in this recalculation. I would like to know how?
We are talking about a special kind of indexing here. This is not sorting; it is just preparing items to end up in a certain order in case someone happens to sort them (by whatever sorting algorithm). I know that variants of this kind of indexing have been used in libraries for decades, maybe centuries, to ensure that books belonging together but lacking a common title end up next to each other in the shelves, but I have never heard of a name for it.
The 8 is probably chosen wisely as a compromise, maybe even by analyzing typical use cases. Consider this: If you choose a small increment, e. g. 1, then all tickets will have ranks like [a, b, c, …]. This will be great if you create a lot of tickets (up to 26) in the correct order because then your rank fields keep small (one letter). But as soon as you move a ticket between two other tickets, you will have to add a letter: [a, b] plus a new ticket between them: [a, an, b]. If you expect to have this a lot, you better leave gaps between the ranks: [a, i, q, …], then an additional ticket can get a single letter as well: [a, e, i, q, …]. But of course if you now create lots of tickets in the correct order right in the beginning, you quickly run out of letters: [a, i, q, y, z, za, zi, zq, …]. The 8 probably is a good value which allows for enough gaps between the tickets without increasing the need for many letters too soon. Keep in mind that other scenarios (maybe not Jira tickets which are created manually) might make other values more reasonable.
You are right, the rank fields get recalculated now and then, Lexorank calls this "balancing". Basically, balancing takes place in one of three occasions: ① The ranks are exhausted (largest value reached), ② the ranks are due to user-reranking of tickets too close together ([a, b, i] and something is supposed to go in between a and b), and ③ a balancing is triggered manually in the management page. (Actually, according to the presentation, Lexorank allows for up to three letter ranks, so "too close together" can be something like aaa and aab but the idea is the same.)
The <bucket> part of the rank is increased during balancing, so a messy [0|a, 0|an, 0|b] can become a nice and clean [1|a, 1|i, 1|q] again. The brownbag presentation about Lexorank (as linked by #dandoen in the comments) mentions a round-robin use of <buckets>, so instead of a constant increment (0→1→2→3→…) a 2 is increased modulo 3, so it will turn back to 0 after the 2 (0→1→2→0→…). When comparing the ranks, the sorting algorithm can consider a 0 "greater" than a 2 (it will not be purely lexicographical then, admitted). If now the balancing algorithm works backwards (reorder the last ticket first), this will keep the sorting order intact all the time. (This is just a side aspect, that's why I keep the explanation small, but if this is interesting, ask, and I will elaborate on this.)
Sidenote: Lexorank also keeps track of minimum and maximum values of the ranks. For the functioning of the algorithm itself, this is not necessary.

n! combinations, how to find best one without killing computer?

I'll get straight to it. I'm working on an web or phone app that is responsible for scheduling. I want students to input courses they took, and I give them possible combinations of courses they should take that fits their requirements.
However, let's say there's 150 courses that fits their requirements and they're looking for 3 courses. That would be 150C3 combinations, right?.
Would it be feasible to run something like this in browser or a mobile device?
First of all you need a smarter algorithm which can prune the search tree. Also, if you are doing this for the same set of courses over and over again, doing the computation on the server would be better, and perhaps precomputing a feasible data structure can reduce the execution time of the queries. For example, you can create a tree where each sub-tree under a node contains nodes that are 'compatible'.
Sounds to me like you're viewing this completely wrong. At most institutions there are 1) curriculum requirements for graduation, and 2) prerequisites for many requirements and electives. This isn't a pure combinatorial problem, it's a dependency tree. For instance, if Course 201, Course 301, and Course 401 are all required for the student's major, higher numbers have the lower numbered ones as prereqs, and the student is a Junior, you should be strongly recommending that Course 201 be taken ASAP.
Yay, mathematics I think I can handle!
If there are 150 courses, and you have to choose 3, then the amount of possibilities are (150*149*148)/(3*2) (correction per jerry), which is certainly better than 150 factorial which is a whole lot more zeros ;)
Now, you really don't want to build an array that size, and you don't have to! All web languages have the idea of randomly choosing an element in an array, so you get an element in an array and request 3 random unique entries from it.
While the potential course combinations is very large, based on your post I see no reason to even attempt to calculate them. This task of random selection of k items from n-sized list is delightfully trivial even for old, slow devices!
Is there any particular reason you'd need to calculate all the potential course combinations, instead of just grab-bagging one random selection as a suggestion? If not, problem solved!
Option 1 (Time\Space costly): let the user on mobile phone browse the list of (150*149*148) possible choices, page by page, the processing is done at the server-side.
Option 2 (Simple): instead of the (150*149*148)-item decision tree, provide a 150-item bag, if he choose one item from the bag, remove it from the bag.
Option 3 (Complex): expand your decision tree (possible choices) using a dependency tree (parent course requires child courses) and the list of course already taken by the student, and his track\level.
As far as I know, most educational systems use the third option, which requires having a profile for the student.

Algorithm for optimizing the order of actions with cooldowns

I can choose from a list of "actions" to perform one once a second. Each action on the list has a numerical value representing how much it's worth, and also a value representing its "cooldown" -- the number of seconds I have to wait before using that action again. The list might look something like this:
Action A has a value of 1 and a cooldown of 2 seconds
Action B has a value of 1.5 and a cooldown of 3 seconds
Action C has a value of 2 and a cooldown of 5 seconds
Action D has a value of 3 and a cooldown of 10 seconds
So in this situation, the order ABA would have a total value of (1+1.5+1) = 3.5, and it would be acceptable because the first use of A happens at 1 second and the final use of A happens at 3 seconds, and then difference between those two is greater than or equal to the cooldown of A, 2 seconds. The order AAB would not work because you'd be doing A only a second apart, less than the cooldown.
My problem is trying to optimize the order in which the actions are used, maximizing the total value over a certain number of actions. Obviously the optimal order if you're only using one action would be to do Action D, resulting in a total value of 3. The maximum value from two actions would come from doing CD or DC, resulting in a total value of 5. It gets more complicated when you do 10 or 20 or 100 total actions. I can't find a way to optimize the order of actions without brute forcing it, which gives it complexity exponential on the total number of actions you want to optimize the order for. That becomes impossible past about 15 total.
So, is there any way to find the optimal time with less complexity? Has this problem ever been researched? I imagine there could be some kind of weighted-graph type algorithm that works on this, but I have no idea how it would work, let alone how to implement it.
Sorry if this is confusing -- it's kind of weird conceptually and I couldn't find a better way to frame it.
EDIT: Here is a proper solution using a highly modified Dijkstra's Algorithm:
Dijkstra's algorithm is used to find the shortest path, given a map (of a Graph Abstract), which is a series of Nodes(usually locations, but for this example let's say they are Actions), which are inter-connected by arcs(in this case, instead of distance, each arc will have a 'value')
Here is the structure in essence.
Graph{//in most implementations these are not Arrays, but Maps. Honestly, for your needs you don't a graph, just nodes and arcs... this is just used to keep track of them.
node[] nodes;
arc[] arcs;
}
Node{//this represents an action
arc[] options;//for this implementation, this will always be a list of all possible Actions to use.
float value;//Action value
}
Arc{
node start;//the last action used
node end;//the action after that
dist=1;//1 second
}
We can use this datatype to make a map of all of the viable options to take to get the optimal solution, based on looking at the end-total of each path. Therefore, the more seconds ahead you look for a pattern, the more likely you are to find a very-optimal path.
Every segment of a road on the map has a distance, which represents it's value, and every stop on the road is a one-second mark, since that is the time to make the decision of where to go (what action to execute) next.
For simplicity's sake, let's say that A and B are the only viable options.
na means no action, because no actions are avaliable.
If you are travelling for 4 seconds(the higher the amount, the better the results) your choices are...
A->na->A->na->A
B->na->na->B->na
A->B->A->na->B
B->A->na->B->A
...
there are more too, but I already know that the optimal path is B->A->na->B->A, because it's value is the highest. So, the established best-pattern for handling this combination of actions is (at least after analyzing it for 4 seconds) B->A->na->B->A
This will actually be quite an easy recursive algorithm.
/*
cur is the current action that you are at, it is a Node. In this example, every other action is seen as a viable option, so it's as if every 'place' on the map has a path going to every other path.
numLeft is the amount of seconds left to run the simulation. The higher the initial value, the more desirable the results.
This won't work as written, but will give you a good idea of how the algorithm works.
*/
function getOptimal(cur,numLeft,path){
if(numLeft==0){
var emptyNode;//let's say, an empty node wiht a value of 0.
return emptyNode;
}
var best=path;
path.add(cur);
for(var i=0;i<cur.options.length;i++){
var opt=cur.options[i];//this is a COPY
if(opt.timeCooled<opt.cooldown){
continue;
}
for(var i2=0;i2<opt.length;i2++){
opt[i2].timeCooled+=1;//everything below this in the loop is as if it is one second ahead
}
var potential=getOptimal(opt[i],numLeft-1,best);
if(getTotal(potential)>getTotal(cur)){best.add(potential);}//if it makes it better, use it! getTotal will sum up the values of an array of nodes(actions)
}
return best;
}
function getOptimalExample(){
log(getOptimal(someNode,4,someEmptyArrayOfNodes));//someNode will be A or B
}
End edit.
I'm a bit confused on the question but...
If you have a limited amount of actions, and that's it, then always pick the action with the most value, unless the cooldown hasn't been met yet.
Sounds like you want something like this (in pseudocode):
function getOptimal(){
var a=[A,B,C,D];//A,B,C, and D are actions
a.sort()//(just pseudocode. Sort the array items by how much value they have.)
var theBest=null;
for(var i=0;i<a.length;++i){//find which action is the most valuable
if(a[i].timeSinceLastUsed<a[i].cooldown){
theBest=a[i];
for(...){//now just loop through, and add time to each OTHER Action for their timeSinceLastUsed...
//...
}//That way, some previously used, but more valuable actions will be freed up again.
break;
}//because a is worth the most, and you can use it now, so why not?
}
}
EDIT: After rereading your problem a bit more, I see that the weighted scheduling algorithm would need to be tweaked to fit your problem statement; in our case we only want to take those overlapping actions out of the set that match the class of the action we selected, and those that start at the same point in time. IE if we select a1, we want to remove a2 and b1 from the set but not b2.
This looks very similar to the weighted scheduling problem which is discussed in depth in this pdf. In essence, the weights are your action's values and the intervals are (starttime,starttime+cooldown). The dynamic programming solution can be memoized which makes it run in O(nlogn) time. The only difficult part will be modifying your problem such that it looks like the weighted interval problem which allows us to then utilize the predetermined solution.
Because your intervals don't have set start and end times (IE you can choose when to start a certain action), I'd suggest enumerating all possible start times for all given actions assuming some set time range, then using these static start/end times with the dynamic programming solution. Assuming you can only start an action on a full second, you could run action A for intervals (0-2,1-3,2-4,...), action B for (0-3,1-4,2-5,...), action C for intervals (0-5,1-6,2-7,...) etc. You can then use union the action's sets to get a problem space that looks like the original weighted interval problem:
|---1---2---3---4---5---6---7---| time
|{--a1--}-----------------------| v=1
|---{--a2---}-------------------| v=1
|-------{--a3---}---------------| v=1
|{----b1----}-------------------| v=1.5
|---{----b2-----}---------------| v=1.5
|-------{----b3-----}-----------| v=1.5
|{--------c1--------}-----------| v=2
|---{--------c2---------}-------| v=2
|-------{-------c3----------}---| v=2
etc...
Always choose the available action worth the most points.

Coming up with factors for a weighted algorithm?

I'm trying to come up with a weighted algorithm for an application. In the application, there is a limited amount of space available for different elements. Once all the space is occupied, the algorithm should choose the best element(s) to remove in order to make space for new elements.
There are different attributes which should affect this decision. For example:
T: Time since last accessed. (It's best to replace something that hasn't been accessed in a while.)
N: Number of times accessed. (It's best to replace something which hasn't been accessed many times.)
R: Number of elements which need to be removed in order to make space for the new element. (It's best to replace the least amount of elements. Ideally this should also take into consideration the T and N attributes of each element being replaced.)
I have 2 problems:
Figuring out how much weight to give each of these attributes.
Figuring out how to calculate the weight for an element.
(1) I realize that coming up with the weight for something like this is very subjective, but I was hoping that there's a standard method or something that can help me in deciding how much weight to give each attribute. For example, I was thinking that one method might be to come up with a set of two sample elements and then manually compare the two and decide which one should ultimately be chosen. Here's an example:
Element A: N = 5, T = 2 hours ago.
Element B: N = 4, T = 10 minutes ago.
In this example, I would probably want A to be the element that is chosen to be replaced since although it was accessed one more time, it hasn't been accessed in a lot of time compared with B. This method seems like it would take a lot of time, and would involve making a lot of tough, subjective decisions. Additionally, it may not be trivial to come up with the resulting weights at the end.
Another method I came up with was to just arbitrarily choose weights for the different attributes and then use the application for a while. If I notice anything obviously wrong with the algorithm, I could then go in and slightly modify the weights. This is basically a "guess and check" method.
Both of these methods don't seem that great and I'm hoping there's a better solution.
(2) Once I do figure out the weight, I'm not sure which way is best to calculate the weight. Should I just add everything? (In these examples, I'm assuming that whichever element has the highest replacementWeight should be the one that's going to be replaced.)
replacementWeight = .4*T - .1*N - 2*R
or multiply everything?
replacementWeight = (T) * (.5*N) * (.1*R)
What about not using constants for the weights? For example, sure "Time" (T) may be important, but once a specific amount of time has passed, it starts not making that much of a difference. Essentially I would lump it all in an "a lot of time has passed" bin. (e.g. even though 8 hours and 7 hours have an hour difference between the two, this difference might not be as significant as the difference between 1 minute and 5 minutes since these two are much more recent.) (Or another example: replacing (R) 1 or 2 elements is fine, but when I start needing to replace 5 or 6, that should be heavily weighted down... therefore it shouldn't be linear.)
replacementWeight = 1/T + sqrt(N) - R*R
Obviously (1) and (2) are closely related, which is why I'm hoping that there's a better way to come up with this sort of algorithm.
What you are describing is the classic problem of choosing a cache replacement policy. Which policy is best for you, depends on your data, but the following usually works well:
First, always store a new object in the cache, evicting the R worst one(s). There is no way to know a priori if an object should be stored or not. If the object is not useful, it will fall out of the cache again soon.
The popular squid cache implements the following cache replacement algorithms:
Least Recently Used (LRU):
replacementKey = -T
Least Frequently Used with Dynamic Aging (LFUDA):
replacementKey = N + C
Greedy-Dual-Size-Frequency (GDSF):
replacementKey = (N/R) + C
C refers to a cache age factor here. C is basically the replacementKey of the item that was evicted last (or zero).
NOTE: The replacementKey is calculated when an object is inserted or accessed, and stored alongside the object. The object with the smallest replacementKey is evicted.
LRU is simple and often good enough. The bigger your cache, the better it performs.
LFUDA and GDSF both are tradeoffs. LFUDA prefers to keep large objects even if they are less popular, under the assumption that one hit to a large object makes up lots of hits for smaller objects. GDSF basically makes the opposite tradeoff, keeping many smaller objects over fewer large objects. From what you write, the latter might be a good fit.
If none of these meet your needs, you can calculate optimal values for T, N and R (and compare different formulas for combining them) by minimizing regret, the difference in performance between your formula and the optimal algorithm, using, for example, Linear regression.
This is a completely subjective issue -- as you yourself point out. And a distinct possibility is that if your test cases consist of pairs (A,B) where you prefer A to B, then you might find that you prefer A to B , B to C but also C over A -- i.e. its not an ordering.
If you are not careful, your function might not exist !
If you can define a scalar function of your input variables, with various parameters for coefficients and exponents, you might be able to estimate said parameters by using regression, but you will need an awful lot of data if you have many parameters.
This is the classical statistician's approach of first reviewing the data to IDENTIFY a model, and then using that model to ESTIMATE a particular realisation of the model. There are large books on this subject.

Resources