rmunro Goto Github PK

active_learning_class

Code for class on Deep Active Learning and Annotation

active_learning_imagenet

Coding exercise to extend tensorflow for imagenet to active learning, for use in a job interview or similar.

annotation_imagenet

Coding exercise to extend to create an annotation tool for ImageNet labels on Machine Learning output

bicycle_detection

Practical example from Human-in-the-Loop Machine Learning book

char-cnn-text-classification-pytorch

Character-level Convolutional Neural Networks for text classification in PyTorch

chichewa

Morphological parser for the Chichewa language

deep_learning_course

disaster_response_messages

This dataset contains 25,000 messages drawn from events including an earthquake in Haiti in 2010, floods in Pakistan in 2010, super-storm Sandy in the U.S.A. in 2012, and news articles spanning a large number of years and 100s of different disasters. The data has been encoded with 38 different categories related to disaster response and has been stripped of messages with sensitive information in their entirety.

dl-dataday-workshop

Deep Learning Course, example code

food_safety

Practical example from Human-in-the-Loop Machine Learning book

gender_augmentation

human-in-the-loop synthetic data generation to augment the universal dependencies corpus with gender-balanced independent possessive pronouns

headlines

Practical example from Human-in-the-Loop Machine Learning book

masked_bias_detection

Code for EMNLP2020 paper: "Detecting Independent Pronoun Bias with Partially-Synthetic DataGeneration"

models

Models built with TensorFlow

pytorch_active_learning

PyTorch Library for Active Learning to accompany Human-in-the-Loop Machine Learning book

sdcnd-behavioral-cloning-p3

Use Deep Learning to Clone Driving Behavior

sms-turks

System to help volunteers manually parse information out of text messages.

uncertainty_sampling_numpy

NumPy implementations of common Active Learning strategies for Uncertainty Sampling

unicode_image_generator

Generates an image for every unicode character