-
Visualize How CNN Makes Predictions
Overview Deep learning model is often a black box. It’s usually hard to understand how the deep neural network learned the features based on which it makes predictions. This blog post will show a technique introduced by Zhou et al[1] that could help us understand what the CNN has learned from the training and based on what it makes predictions. …
-
Transfer Learning with Keras
Overview Deep learning with Deep Neural Network has gained huge momentum in the past few years owning to the blooming of data and GPU accelerated computation. The former makes it possible to collect massive amount of training data to tune millions or even billions of parameters; the later makes running deep learning computation approachable for everyone have access to a consumer PC with GPU installed. …
-
Anomaly Dection
A Generic Approach to System/Data Monitoring …