Course Outline
During this course the following arguments will be explained:
- Basics of Artificial Neural Networks (ANN)
- Deep learning current Approaches in Python & Tensorflow
- Data preparation for ANN processing
- Applications of ANN on real world dataset
- Convolutional and Recurrent Neural Networks (CNN & RNN)
By the end of the course participants will be able to:
- Read, upload and prepare data for ANN processing
- Create a neural network
- Evaluation of a Neural Network, modify and optimize the efficiency
- Apply different approaches of ANN (CNN & RNN) on various cases
Course Instructor
Reza Paya
Software Engineer, Experienced CTO/CIO with a demonstrated history of working in the computer software industry. Highly Skilled in Analytical systems, data semantics and integration, information visualization, augmented reality and IoT.
• Alumnus of innovactionLab.
• Community leader and mentor in Startup weekend Rome.
Selected as a successful foreigner in HITECH startup in Italy by FWD.us in 2016.
• Winner of “MoneyGram Award 2016” for the Foreign entrepreneur in innovation section.
• Speaker at international Information Architecture day, Rome/2016.
• Big Data Course Coordinator/Tutor Geeks Academy.
Course Program:
Artificial Neural Networks (theory)
• Intro to ANN. Perceptron as a biology neuron.
• Mathematical model.
• Perceptron vs. Logistic Regression.
• Feedforward Neural Networks.
• Activation Functions.
• Cost Functions. Binary and multiclass Cross Entropy.
• Gradient Descent.
• Training – Backpropagation
• Overfitting
• Model evaluation
Implementing ANN with Tensorflow (hands-on)
• Intro to Tensorflow and Keras
• Keras syntax – Data preparation
• Keras syntax – Creating and training a model
• Keras Model evaluation
• Binary Classification example
• Multiclass Classification example
• Regression example
• Project 1 – Classification
• Project 2 – Regression
Convolutional Neural Networks (theory and hands-on)
• Images processing. Use cases.
• Convolution operation. Filters and kernels.
• Convolutional layers in Tensorflow.
• Pooling layers. Dropout concept.
• Batch normalization.
• Implementing CNN with Tensorflow.
• CNN laboratory. Images classification.
• Transfer Learning. Using pre-trained CNN.
Recurrent Neural Networks (theory and hands-on)
• RNN basic theory. Memory neurons.
• Time series data. Seasonality and trends.
• Basic analysis for Time Series. Examples.
• Basic RNN implementation. Text processing example.
• Issues with short term memory cells.
• Vanishing gradient.
• LSTM (Long Short Term Memory) cells theory.
• Implementing a LSTM example in Tensorflow.
• LSTM laboratory using a real dataset.
• Natural Language Processing example.