Tentative Schedule

Date	Instructor	Title
Day 1 (November 16)/5-5:30	Pascal	BigData, Machine Learning and text mining by example. Using PySpark (Spark+Python). Why Spark for BigData and ML: Basic knowledge for Text Mining:LSA,DSA,TF-IDF,Word2Vec. installing pyspark on Windows or Linux and first basic examples
Day 2 (November 17)/5-8	Pascal	Implementing 1 or 2 relalistics Text Mining examples
Day 3 (November 19)/5-8	Hikmat	Introduction, Logistic Regression (Identifying Shipts), Shallow Network (Movie Reviews), Backpropagation
Day 4 (November 21)/5-8	Hikmat	Feedforward Networks (MNIST digits), Tensorflow and Keras
Day 5 (November 24)/5-8	Hikmat	Convolution Networks (MNIST digits), Recurrent Networks
Day 6 (November 26)/5-8	Hikmat	Autoencoders (denoising), Variational Encoders(data generation), Generative Adversarial Networks
Day 7 (November 30)/5-8	Walid	Machine learning application:Face Mask Detection
Day 8 (December 1)/5-8	M. Hülsman	Introduction to NLP
Day 9 (December 2)/5-8	Walid	Title 2
Day 10 (December 3)/5-8	Herpers	CNN applications

Neural Network and Text Mining

Neural Networks

This course introduces Neural Network for Machine Learning. It is meant to be a hands-on course without sacrificing the conceptual background. As such the first few code examples are developed (using numpy or cupy) from basic principles without using any ML framework. After that we use Tensorflow v2 for almost all examples.

Already done

Below are already done (the code) but need (a lot of) polishing. Especially integrating the lecture with the code.

1. Introduction
2. Logistic Regression: basic concepts, perceptron, activation function, loss and trainning. Two different applications: one using CIFAR10 data set and the second using sentiment analysis in Tweets
3. Neural Network with a single layer: a single hidden layer, backpropagation example, training using gradient descent. Using a generate data set
4. The same example as the previous but using Tensorflow low level API (as opposed to Keras) automatic differentiation.
5. Feedforward Networks: design of a feedforward network with an arbitrary number of layers and backpropagation from first principles using numpy (cupy) with no framework used. Application to MNIST data set
6. Same as the above but using the high level Keras API in tensorflow
7. Convolution Networks using Keras. Application to CIFAR10 data set

coming soon

1. Recurrent Neural Networks: using RNN in Tensorflow to detect spam messages
2. Autoencoder.

Text mining

Visit this directory BigData