Three Days IEEE Skill Development Hand’s on Workshop on “Machine Learning & Deep Learning”

IEEE student’s branch, DBCE organized a three days Skill Development Hands-on workshop on “Machine learning and Deep learning” in association with IEEE Bombay section for the students of third year and final year ETC and Computer department students & faculty members on 23, 24 and 25 August 2019 coordinated by IEEE Branch Counselor, Prof. Yeshudas Muttu. The resource person for the workshop was Prof. Santosh Chapaneri from St. Francis Institute of Technology, Mumbai, recognized & registered to conduct IEEE Skill Development workshops under IEEE Bombay Section. The main objective of this workshop was to provide an overview of Python programming language & provide a hands-on experience on Machine learning and Deep learning.

Day 1: August 23, 2019

IEEE Student Member, Ms Krystal Fernandes welcomed everyone for the three days’ workshop, introduced the speaker to the audience & requested him to take over the session.

The Session began with a wide overview of Python programming language which included basic arithmetic calculations, importing libraries in python. Dictionary in python, strings, lists and various data types was introduced to the students & short exercises on the topics that were covered so far were given. Various programming loops in python were summarized to the students & various functions along with the loops were introduced. Important tools in python like ‘numpy’ & – ‘pandas’ were introduced to the attendees. Numpy was further used to implement various activation functions like sigmodal, ReLU & Softmax. The modules contained in script for optimization, linear algebra, integration, interpolation, special functions, FFT, signal and image processing were taught. Students were then given a task to perform IMDB data analysis using ‘pandas’.

The second half of the day began with the basic concepts of machine learning. The prerequisites of machine learning told to the students were linear algebra, probability, statistics and calculus. Noel complexity (Bias variance trade-off), underfitting and overfitting complexities, curse of dimensionality, Dataset split, confusion matrix, Gradient descent algorithm. To conclude it for the day, Logistic regression & linear regression were covered.

Day 2: August 24, 2019

Prof. Santosh Chapaneri began the session with a brief review of classification and regression in Machine learning. The topic of logical regression was further carried by coding it on google colab notebook. IRIS dataset was used to extend binary classifier to solve multiclass problems. Students were made to solve linear regression problem and later executed on the computer for better understanding. Naive Bayes, a supervised classification technique was discussed, with the knowledge of Bayes theorem and conditional independence being the basic requirement. Later, the program was executed and formula for feature extraction from text data using TFIDF was discussed. The sessions were well balanced between theoretical & practical aspects.

Post lunch, the session began with Principal Component Analysis (PCA) algorithm. This was later applied to the dataset using google colab that provided scattered data graph. The final topic of the day was clustering & its significance. Starting off with mathematical analysis, collections, CH index and its drawbacks, K-means clustering was implemented. To conclude, the speaker encouraged the students to join various web-based organizations based on machine learning and deep learning.

Day 3: August 25, 2019

The day started with an introduction to Deep Learning, which is a subfield of Machine Learning. Prof. Santosh Chapaneri explained the significance of Deep Learning and compared to Machine Learning as a whole. The MNIST dataset was used in the Deep learning demonstrations. Later, the session progressed by teaching the concepts of ‘Perceptron’ and ‘Neural Networks’. In this, various neural networks were taught, starting with Artificial Neural Network. The concept of layers, neurons, activation functions and the parameters involved in a given neural network were efficiently explained. The other neural networks taught to the students were the Deep Neural Network and the Convolutional Neural Network. Examples on each neural network were solved, and Backpropagation algorithm was demonstrated. Deep Learning libraries were introduced to the students of which, TensorFlow and Keras were used in the practical implementation of Deep Learning concepts in the latter part of the day.

In the afternoon session, an efficient neural network algorithm was used to work on the MNIST dataset. The APIs from TensorFlow and Keras libraries were used. The concept of Convolutions and Max pooling was implemented. As the end of the session neared, test architecture was given to the students to be implemented with the help of concepts learned during the workshop. The students successfully completed the task. Finally, the resource person answered questions and gave valuable advice to the students to delve further into this field. A vote of thanks was given by the Chairperson of IEEE-DBCE Mr. Sherwin Colaco, and a token of appreciation was presented by IEEE Branch Counselor, Prof. Yeshudas Muttu to the resource person. Overall positive feedback was received by the students towards this workshop.