Kyle Kastner

deform-conv. Deformable Convolution in TensorFlow / Keras

PyCon2015. Material for talk "Machine Learning 101" https://speakerdeck.com/kastnerkyle/pycon2015 https://us.pycon.org/2015/schedule/presentation/367/

kaggle-dogs-vs-cats. Code for Kaggle Dovs vs. Cats competition http://www.kaggle.com/c/dogs-vs-cats

tools. Various tools for graphs, audio, images

representation_mixing. Demos, pretrained models, and (WIP) code supporting Representation Mixing

SciPy2015. Talk for SciPy2015 "Deep Learning: Tips From The Road"

kastnerkyle.github.io. Github blog

multi-speaker-tacotron-tensorflow. Multi-speaker Tacotron in TensorFlow. 오픈소스 딥러닝 다중 화자 음성 합성 엔진.

diphone_synthesizer. A tutorial diphone synthesizer in Python

kaggle-cifar10. Code for Kaggle Cifar10 competition http://www.kaggle.com/c/cifar-10

kaggle-criteo. Code for Criteo competition http://www.kaggle.com/c/criteo-display-ad-challenge

SciPy2013. Slides and code for SciPy 2013 talk - "A Gentle Introduction To Machine Learning"

net. Neural networks in Theano (ABANDONED/DISCONTINUED) - see dagbldr for a continuation of this code with some new tricks

harmonic_recomposition_workshop. Python

raw_voice_cleanup. Examples of cleaning up raw voices

ez-phones. Wrapper to pocketsphinx phoneme labeling tools

pachet_experiments. Python

pytorch-text-vae. Python

research_megarepo. A monster repo for random research, not organized in any particular way

g2p_pi. Simple rules based grapheme to phoneme in Python

romanian_noise_tests. Samples and visualization of a small test for speech synthesis - see the paper at http://www.josesotelo.com/speechsynthesis/

quadcopter-brain. Repository for work on a quadcopter brain using hacked wifi SD cards

magenta. Magenta: Music and Art Generation with Machine Intelligence

PyTexas2013. Code and slides for PyTexas 2013 talk, "Trends in Deep Learning"

speech_density. Speech modeling using code by Kratarth Goel http://dblp.uni-trier.de/pers/hd/g/Goel:Kratarth

image-analogies. Generate image analogies using neural matching and blending.

udem_masters_thesis. My (English, after the abstract) thesis, "Structured Prediction and Generative Modeling using Neural Networks". Thanks to my professors (Aaron Courville, Roland Memisevic, Yoshua Bengio) and coauthors (Junyoung Chung, Laurent Dinh, Amjad Almahairi, Francesco Visin, Kyunghyun Cho, Kratarth Goel, and others)

analysis_of_audio_superresolution_using_neural_nets. Analyzing samples from ICLR workshop paper https://openreview.net/forum?id=S1gNakBFx&noteId=S1gNakBFx

WaveRNN-1. A WaveRNN implementation

arrayprocessing. Array processing algorithms and tools for RF array design, direction of arrival (DOA, also known as DF) estimation, and geolocation.

deviation. Deviation for the Devo F7

santa_barbaria. Standalone experiments

PyGotham2015. Slides for PyGotham2015

vrnn-samples. Audio samples from "A Recurrent Latent Variable Model for Sequential Data"; J. Chung, K. Kastner, L. Dinh, K. Goel, A. Courville, Y. Bengio http://arxiv.org/abs/1506.02216

pylearn2. A Machine Learning library based on Theano

rmelnet. Experimental dump of R-MelNet related code and demo files

crikey. Research experiments

pyrnn. Python bindings for RNNLIB

deconstructionism. deconstructing a working model from https://github.com/Grzego/handwriting-generation

Sandbox. Toy implementations of things which may (or may not) grow into something else

tfbldr. Tensorflow models and tools for research

EuroScipy2014. Slides and small scripts for EuroScipy talk "Neural Networks for Computer Vision". Abstract can be found here: https://www.euroscipy.org/2014/schedule/presentation/2/

hmm_tts_build. a direct repository for building and using a "simple" tts

scipy_2015_sklearn_tutorial. Scikit-Learn tutorial material for Scipy 2015

keras-dcgan. Keras implementation of Deep Convolutional Generative Adversarial Networks

Configs. Configs for different programs

fruitspeech_dataset. Github host for Hakon Sandsmark's Fruitspeech Dataset

Variational-Recurrent-Autoencoder. Implementation of the VRAE

sparse_attention. Examples of using sparse attention, as in "Generating Long Sequences with Sparse Transformers"

jukebox. Code for "Jukebox: A Generative Model for Music"

makeMoE. From scratch implementation of a sparse mixture of experts language model inspired by Andrej Karpathy's makemore :)