ContentsMayBeHot

At the time of my last blog post, we were managing quite a few problems. Our model was essentially vaporware, our training and testing was hindered by slow, blocking function calls from our loader, and our VRAM was continually getting exhausted during training sessions. But there is nothing to worry about. We have made major strides since then. Major strides. Model improvements First, we have completely overhauled our model's architecture. We are now using a model composed of special layers that combine the functionality of a 2D convolutional neural network with that of an LSTM. Here is a summary of  our model as printed by Keras: This model was made with the help of the wonderful community over on Stack Overflow . I would also like to mention that Professor Auerbach made invaluable contributions. In general, his tutelage made this project possible. We dropped our Sequence subclass, and replaced it with training and testing loops. In these loops, we iterate over the whol

After our last meeting, Professor Auerbach asked us to shift our focus towards building and training our model. So that's what we've been working on lately. The results so far have been interesting and problematic. The first step was to define a minimal working model and a loading system to feed it our labelled data-set. I wrote a Sequence subclass, which is essentially a kind of generator designed for use with the fit_generator method. With fit_generator and a sequence, we're able to train and test the model with just a couple of one-liners: model.fit_generator(sequence) model.evaluate_generator(sequence) The sequence subclass also has a few other tricks up its proverbial sleeve. For one, it reduces the dimensionality of the frame buffer data from 135×240×3 to 135×240×1 by converting it to gray-scale. This reduces the number of features from 97,200 to 32,400. For two, it does the same with the labels, combining and dropping 26 action types into just 9 atomic clas

replaymanager.py is only 339 lines long but it feels at least three times that when I'm working on it. The module is definitely due for a refactor. For one, the term subdataset should be renamed to version_set and extracted into a class. version_set more accurately and describes what it is, and the class extraction would clean up the namespace in ReplayManager. There is probably some kind of class extraction possible for replays, so that their names, paths, file streams, and Twitter profiles can all be neatly encapsulated, thereby cleaning up the namespace even more. However, I do not want to worry about having two kinds of replays: the one used by ReplayManager and the one used by ReplayParser. Even though ReplayManager does not use ReplayParser, the prospect of making things more muddier deters me. There's probably a right way to refactor this code, but, to put it simply, now is not the time. Speaking of time, I came up with a great way to get work done, even when I am slee

This most recent work period involved a lot of refactoring and adding some new key functionality. Matthew asked me to create a simplified Action Type in addition to the one that was all in place, basically just the same actions without PRESSED and RELEASED. Since we still wanted the original structure to be there, all I had to do was cast the "complex" actions to "simple actions. Matthew then asked if I could convert that SimpleAction type into a matrix, so we could have a clearly defined Y. This was also incredibly easy. I am actually quite happy with how it works as well. All you have to do to create an array for the action is two steps! matrix = numpy.zeros(26) if action is not SimpleAction.INVALID:     matrix[action] = 1; The 26 is the number of different Simple Actions we have. Then, to make it so we can run the parser separately from the Agent, I made it the replay can output numpy files for each character where each row in the file contains the frame of

Since the beginning of the semester, Matthew and I have been woking hard on the Replay Parser and Frame Collector. Both of these parts are crucial to the success of our project. With the first Demo of our project done also comes the completion of this first milestone. Parsing the replay exposed some interesting information to us about how inputs are recorded, and therefore how the game sees them. Our original understanding of inputs was that multiple inputs would actually be spread out across multiple frames, however this was not true. It turns out that a frame is followed by a list of events which take place on that frame. We also learned that some human actions translate into multiple inputs. For example: If you are using the control stick to move your character and you hold right on the control stick, your character moves right. One may think that this means the action RIGHT_PRESS is input, in actuality a list of actions is generated. The list may look something l

Since January, we've been working hard to not only finish writing the Replay Parser and Frame Collector but also totally synchronize them. I'm pleased to report our success. This is an amazing milestone for us because it means that we've surmounted one of our most troubling obstacles. I have also made sure to keep our documentation up to date. So, if you like, you can follow along with this blog post by replicating its results. The Frame Collector uses timed input sequences to start each replay associated with the currently running game version. Then, after waiting a set amount of time for playback to begin, it starts grabbing 1/4-scale frames at a rate of 10 frames per second. The Frame Collector takes these down-scaled frames, which are NumPy arrays, and rapidly pickles and dumps them into the file system. Here's a screenshot of the Frame Collector in action: If you look at the image above, you'll see that each pickle (the .np files)