A state of the art deep neural network archetecture for question/answering. Uses multiple RNN's, episodic memory, and natural language processing.
An overview of CNNs and a comparison of different models with the MNIST and CIFAR10 datasets.
Introductory data science notebook. Includes data cleaning, feature engineering, data visualization, and analysis.
Takes a database of faces and uses vector space manipulation to identify key facial features and identify the face in a new picture.
Explores graphs and connectedness in python, scrapes iMDB for actors and their movies, and then uses breadth first search techniques to find the shortes path connecting actors to Kevin Bacon.
Using Markov chains for forcasting wheather and language prediciton. The final implementation is a Twitter-Bot that scrapes the web for song lyrics and produces and tweets randomly generated sentences.
Uses eigenvalues and graph theory to analyze images. These concepts are then implemented by a two tone image segmentation.
Explains the theory and reasoning behind finding the nearest multi-dimensional neighbor with a K-D Tree. This method is then used for handwriting recognition.
Explains the basics of digital signals and proccessing with the Fourier transform.
Uses the Fourier transform to filter out unwanted noise and convolve signals to reproduce accoustics.
Explores the concept behind the Singular Value Decomposition and then uses it for image compression.
Experiments with basic data visualization in python with matplotlib.