NCI Courses

Module Introduction; History and evolution of programming languages used for AI; Programming types and paradigms (imperative, declarative, functional, logic, agent oriented programming, probabilistic programming, etc.); Overview of programming languages used for implementing AI solutions: general-purpose languages (e.g., Python), classical AI languages (e.g., Lisp), new generic AI programming languages (e.g., MIT Gen), deep probabilistic programming languages (e.g., Edward, Pyro).

Challenges associated with big data requirements of statistical AI (e.g., deep learning); Computation challenges (e.g., search space, time and space complexity); Parallelism for computational processes; Use of specialised/dedicated hardware to speed up computations (e.g., GPUs, Google TPUs, wafer-scale AI chips such as Cerebras CS-1, etc.); Distributed computing platforms; Brief overview AI services and APIs offered by public cloud providers (e.g., Amazon AWS, Microsoft Azure, Google Cloud Platform).

Syntax and semantics, expressions and statements; Basic data types, conversion and coercion; Built in data structures (arrays, matrices, lists, etc.), indexing data structures; Program flow control and iteration.

Input/output data from structured/semi-structured file formats (csv, xml, json); Input data from the Internet (e.g., web scraping, web APIs); Defining functions; Lambdas for functional programming; Algorithm design.

Dealing with missing values; Catching exceptions; Use of support libraries (e.g., Pandas, Numpy, dfply); Stream input sources, live data stream, window-based transformations, combination of batch and stream computations.

Database system technologies; Programmatically connecting to databases; Create/Read/Update/Delete (CRUD) Operations; SQL Optimization, Indexing and Normalization.

NoSQL Overview and Data Models: Document Model, Key-Value Model, Column Family, Aggregates, Graph Model, Triple Stores; NoSQL Data Modelling Concepts; Query Languages for Data in NoSQL; NoSQL systems.

Data wrangling techniques; Developing programs for data processing activities (e.g., data extraction, cleaning, merging, aggregation, validation, analysis, reporting).

Ontology definitions: domain ontology, concepts, instances and relationships. Overview of technologies for ontology engineering: Web Languages (e.g., HTML, XML and RDF), Metadata standards (e.g., Dublin Core), Ontology Language (e.g., OWL), Ontology Editor (e.g., Protégé), Reasoning language (e.g., SPARQL), reasoners (e.g., HermiT); Overview of Python packages for ontology-oriented programming (e.g., Owlready2, RDFlib, pyspaql, pronto, AllegroGraph).

A brief introduction to deep learning concepts; Overview of deep learning frameworks (e.g., PyTorch, TensorFlow, Apache MXNet, Keras); Overview of public cloud AI services for deep learning (e.g., AWS Deep Learning AMIs, Google Cloud TPUs); Use of pre-trained models and cloud services for various example applications (e.g., regression, classification).

Overview of NLP libraries and frameworks (e.g., NLTK); Overview of public cloud AI services for NLP, translation (e.g., Amazon Lex, Polly, etc.); Use of pre-trained Generalized Language Models for NLP applications (e.g., Google BERT, OpenAI GPT-2, etc.).

Overview of image processing libraries and frameworks (e.g., OpenCV); Overview of public cloud AI services for image and video recognition (e.g., Azure Face, AWS Rekognition, etc.); Use of pre-trained models for example applications (e.g., RetinaNET object detection).