Data Science (Conversion) MSc

This module will provide students opportunity to understand and apply computational techniques to analyse and synthesize natural language and speech – Natural Language Processing (NLP). An interdisciplinary bridging of linguistics, information retrieval and machine learning will provide necessary skills to develop applications capable of comprehending, manipulating and generating natural language text and speech similar to Large Language Models.

This module will introduce topics in NLP including tokenization, stemming, parsing, lemmatization, basic text processing, linguistics and NLP tasks, Python NLTK library for NLP, text preprocessing and n-grams, Softmax / MAXENT (sequence) classifiers, sequence

classifiers for POS and NER, Deep learning-based word representations & deep networks

for NER, recurrent networks and language modelling, statistical machine

translation, word alignment, parallel corpora, decoding, evaluation, modern deep learning machine translation systems (phrase-based, syntactic), syntax and parsing, co-reference resolution, tree recursive neural networks for POS tagging, computational semantics, question answering, text summarization and dialogue systems.

High-Performance Computing (HPC) aspects will demonstrate how NLP can be leveraged on graphical processing units (GPUs) using Google TensorFlow and NLTK library. Focus is primarily upon the application of NLP to real-world problems, with some introduction to transformers and large language models, like ChatGPT, with practical exercises using.

This module develops the theoretical, mathematical and practical foundations of algorithms in Computer Science. The time and space trade-offs and their relation to size and nature of inputs are fundamental in all software applications of programming, databases and distributed computing, machine learning, computer vision, deep learning, natural language processing, big data analytics, cryptography and information retrieval etc.

The module aims to introduce students to an in-depth understanding of ‘best’, ‘average’ and ‘worst’ case scenarios, iteration versus recursion, backtracking, linear versus dynamic programming, greedy algorithms, self-balancing trees, topological graphs (directed and undirected, traversals, colouring, distance algorithms, spanning trees), sorting, hashing and searching algorithms. Topics covered include analysis on nature and size of inputs, asymptotic notations (Big-O, Big Ω, Big Θ, little-o, little-ω), recursion and recurrence relations, design of algorithms: brute force, divide and conquer and greedy approach, dynamic programming; elements of dynamic programming, search trees; heaps; hashing; graph algorithms, shortest paths, sparse graphs, string matching, polynomial and matrix calculations and complexity classes.

A variety of algorithms are practically implemented using Python programming language (with open-source libraries) so that upon completion of the module, students should be able to critically explain the mathematical concepts and apply algorithms appropriate to a particular situation.

This module develops the mathematical and statistical underpinnings of artificial intelligence and data science domains. These fundamental concepts are used for analysis of different datasets for forecasting the values, predicting the unknowns, relating the variables for getting deeper insights and indicating data differences with real world complexities.

Topics will cover following areas:

Introduction to statistics and probability, statistical inference, samples, populations, sampling procedures, discrete and continuous variables, types of statistical studies, sample space, events, conditional probability, independent and identically distributed data, product rule, Bayesian inference, statistical moments, variance and covariance of random variables, discrete and continuous probability distributions, central limit theorem, t-distribution, f-quantile and probability plots, single sample & one-and two-sample tests of hypotheses. the use of p-values for decision making in testing hypotheses, linear regression and correlation, least squares, linear regression model using matrices.

Linear algebra: vector spaces, projections, linear transformations, singular value decomposition, PCA and eigen decomposition, power method.

Optimization: Matrix calculus with Lagrange Multipliers, gradient descent, coordinate descent, introduction to convex optimization.

Students will gain knowledge and hands-on experience using Python programming language by implementing specific algorithms for extraction and selection of statistical features, data curation, interpolation and extrapolation, kernel methods, probabilistic reasoning and graphical models, likelihood estimations, dimensionality reduction, principal components, discriminant analyses, singular value decomposition, auto-regression, moving averages, penalised cost functions, generalization, regularization and inference.