This repo is for storing miscellaneous code that I've written.
My implementations of model-free and model-based RL methods. The model is DKF/DVAE.
Experiments involving VAEs. I compared VAE to IWAE. I examined the how well VAEs could model multimodal distributions. I looked how the structure of the VAE affects its performance.
Early implementation of the DKF/DVAE, which is a VAE model for sequential data.
My implementations of various sampling methods.
My implementation of Normalizing Flows. Didnt not work too well.
Autograd examples of SVI
Importance Weighted Mixture of Gaussians Autoencoder
My implementations of various methods including Gibbs sampling, Collapsed Gibbs, and DPGMMs.
Analyzed how to deal with datasets that had unbalanced classes in an online and unsupervised setting.
###Hockey_Predictions
Some code I used to make predictions for drafting in the hockey pool.
Database Course Project - University of Toronto Argo is an automated mapping layer that runs on top of a traditional RDBMS, and presents the JSON data model directly to the application/user. Argo has been previously shown to have a significant speedup over document-oriented databases, specifically MongoDB. This was shown by evaluating the systems using the benchmark NoBench. This benchmark evaluates queries on relatively simple JSON documents. Therefore, in order to gain a more comprehensive assessment of Argo, I extended the NoBench data to include more complex data, such as highly nested objects and arrays with more elements. Here I show that increasing the complexity of the data reduces the speedup that Argo has over MongoDB.
Code I used to make plots in Python using Matplotlib.
Python code to send and receive messages (packets) Used during the summer of 2014 to test packet throughput
Plan to write code to vizulize neural net training.
McGill Course
The code I wrote for some of the assignments in this course. (Java)
McGill Course
The code I wrote for some of the assignments in this course. (Java)
Examples on how to use the Python multiprocessing packages
Neural Net code that I modified from Michael Nielson
Basic examples to help me gain some intuition about regression
Samples from Gaussian distribution and Uniform distribution. Choose number of features and samples. Choose number of features that are distinct between classes.