Course Content

Course Code Core Courses

AI6101 Introduction to AI and AI Ethics

AI6102 Machine Learning: Methodologies and Applications

AI6103 Deep Learning and Applications

AI6104 Mathematics for AI

Bridging Course

AI6120 Python Programming

Elective Courses

AI6121 Computer Vision

AI6122 Text Data Management and Processing

AI6123 Time Series Analysis

AI6124 Neuro Evolution and Fuzzy Intelligence

AI6125 Multi-Agent System

AI6126 Advanced Computer Vision

AI6127 Deep Neural Networks for Natural Language Processing

AI6128 Urban Computing

AI6129 AI Master Project

*Note: Not all courses listed in the curriculum will be offered in a semester.
Courses offered are subject to availability of instructors and resources.
AI6101 Introduction to AI and AI Ethics

This course introduces the basics of artificial intelligence (AI), paying special
attention to the ethical issues involved. It offers an interdisciplinary overview of AI to
help students start their learning journey in this field. Well established paradigms
pertaining to the formal representations of artificial intelligence and knowledge
engineering for computational problem solving will be introduced.

The main topics covered include:

1. The foundation of AI, history, and recent progress
2. Search and intelligent agents
3. Reinforcement Learning
4. Computational game theory
5. The philosophical foundations of AI ethics
6. Techniques for incorporating ethical considerations into AI systems
7. Example applications of ethical AI – A

AI6102 Machine Learning: Methodologies and Applications

* Prerequisite: pass AI6104 (except those with who have Math degrees / taken
probability and linear algebra courses (full semester courses)

This course aims to provide an introductory but broad perspective of machine

learning fundamental methodologies, and show how to apply machine learning
techniques to real-world applications. It is relevant for anyone pursuing a career in AI
or Data Science.

The teaching content of this course includes different machine learning

methodologies in various machine learning paradigms, such as supervised learning,
semi-supervised learning, unsupervised learning, etc., as well as applications of
machine learning.

AI6103 Deep Learning and Applications

Deep learning has recently introduced a paradigm shift from human-design features
to end-to-end systems, and has revolutionized several fields including computer
vision, speech recognition, and natural language processing. Top IT companies like
Google, Facebook, Microsoft, Apple, Amazon have been actively redesigned their
products with deep learning techniques, and the impacts in the coming decades will
go beyond self-driving cars, strategic games like Go, and MRI cancer detection.

The main objective of this course is to introduce the mathematical foundations, the
state-of-the-art architectures, and a professional library of deep learning
architectures. Students will learn how to design their own artificial neural network to
solve their data analysis task. They will also learn how to code efficiently these new
algorithms using PyTorch, one of the most powerful libraries in this field.

AI6104 Mathematics for Artificial Intelligence

This course aims to provide the appropriate mathematical background to students

who will study other courses in the Master of Science in Artificial Intelligence
programme. Upon completion of this subject, the student should be able to:

• Solve systems of linear equations

• Use linear independence basis to decompose vectors.
• Compute partial derivatives
• Perform approximation based on Taylor series
• Search maximum and minimum in multivariable functions.

AI6120 Python Programming

This course aims to provide appropriate computing background to students who will
study other courses in Data Science and AI. Upon completion of this subject, the
student should be able to:

• Use the Jupyter Notebook environment for Python.

• Solve standard algorithmic problems in Python.
• Perform standard numeric computations in Python.
• Handle data in Pandas, and graphics in Matplotlib.
•

AI6121 Computer Vision

Computer vision has been attracting increasing interests thanks to the recent advances in
deep learning especially in convolutional neural networks, recurrent neural networks, etc.
Top IT companies around the world such as Google, Apple, Alibaba, Amazon, Tencent and
Baidu have invested and will continue to invest heavily in various computer vision
technologies due to demands in a wide spectrum of applications and domains such as
robotics, autonomous driving, surveillance and security, computer-aided medical diagnosis,
etc.

The main objective of this course is to introduce the relevant mathematical foundations for
computer vision and machine learning, the basic image analytics and machine learning
technology in computer vision, and the computer vision technologies in detection,
recognition and classification tasks. A series of real-world problems and challenges will be
presented throughout the course via case studies and projects. Students will have a good
understanding of various machine learning and computer vision technologies at the end of
this course. They will also learn how to design their own machine learning and computer
vision systems for various real-world problems.
•

AI6122: Text Data Management and Processing

This course covers fundamental techniques to manage and process text data. The
main topics include: 1) text indexing and search: inverted index, query processing,
ranking, and evaluation, 2) word-level, sentence-level, document-level, and
collection-level processing: morphological analysis, part-of-speech tagging, parsing,
summarization, classification and clustering, and topic modeling, and 3) case studies
and applications: social media text, sentiment analysis, and information extraction.

AI6123: Time Series Analysis

Many of the complex systems are dynamic systems in which their states change
over time. This course introduces time series models and the corresponding
methods for data analysis and inference. Topics include regression models,
autoregressive (AR), moving average (MA), ARMA, and ARIMA processes,
stationary and non- stationary processes, seasonal processes, identification of
models, estimation of parameters, diagnostic checking of fitted models, rare event
detection, forecasting, spectral analysis and time series models of
heteroscedasticity . Real world applications for understanding characteristics,
modelling and evaluating forecasts of time series data in economics, finance and
industries are elaborated with lab on using R.

AI6124 Neuro Evolution and Fuzzy Intelligence

We begin with a short introduction of learning, adaptation andcognition: examining: learning

forms, supervised, unsupervised, reinforcement; Adaptation mechanism - error driven,
performance objective; cognitive knowledge representation and reasoning for rational
decision making. The tools for performing such learning via evolutionary and neural
computations are examined. Cognitive organisation of knowledge and the integration of such
structures with soft computing models are also discussed. Introduction to knowledge
abstraction and expression are discussed in relation to the evolving deep neural structures
that can be used to evolve and to directly learn such cognitive knowledge structure from
data. Symbolic and semantic (Fuzzy representation) knowledge are considered as sound
deductive inference systems. Their logical representation and philosophy within the cognitive
knowledge organisation are considered. Fuzzy logic and fuzzy inferences are covered. Their
implementations within a deep learning neural structure is discussed. Different form of
learning neural network that can be dynamically evolve or construct the fuzzy rules on the
basis of data is covered. We will also cover model cross validation and verification for the
assessment of the quality of the constructed/evolved fuzzy neural networks. Issues of
concept formation, rule association, data drifts and shifts, forgetting, positive and negative
potentiation, evolutionary trade-off between expression and abstraction will be covered.

Next, we examine related cognitive intelligence applications such as those: in computational

finance - risk based trading system, asset assessment, fuzzy trading models; Computational
Attention focus - emotion, sentiment and motivation modelling.

AI6125 Multi-Agent System

This course covers an introduction to multi-agent systems. It will begin by introducing

multi-agent systems and a number of interesting application areas, as well as explain
the complexity of a definition. Decision making and planning in these multi-agent
systems are then addressed. Multi-agent systems are also explained through the
concept of conflicting interactions, game theory and the equilibrium, categorized as
the systems of benevolent agents and those of self-interested agents. In each topic,
students are given several state-of-the-art papers in the literature to read and
present. The course will have a large project component. Students are asked to
implement their own multi-agent system and present their project outcome in the
end.

AI6126 Advanced Computer Vision

The rise of deep learning leads to a wave of breakthroughs in computer vision,

opening up a number of possibilities that are unimaginable before. A lot of new
applications emerge, including face recognition, image super-resolution, and
machine vision in self-driving cars.

This course aims to develop specific knowledge in advanced computer vision by

using deep learning techniques. The course will cover both image and video
recognition, including image classification, various object detection techniques,
image and video segmentation, image super-resolution, and video captioning. Focus
will be on both conceptual understanding of techniques, as well as on practical
implementation of these techniques using modern deep learning frameworks.
Students will also gain experience in critically reviewing scientific literature relevant
to the area. Course project will encourage students to solve real-world research
problems, and also give them experience in written and oral scientific
communication. Overall, this course will provide good exposure to students who
intend to conduct research & development in computer vision.

AI6127: Deep Neural Networks for Natural Language

Processing

Natural language processing (NLP) is one of the most important fields in artificial intelligence
(AI). It has become very crucial in the information age because most of the information is in
the form of unstructured text. NLP technologies are applied everywhere as people
communicate mostly in language; language translation, web search, customer support,
emails, forums, advertisement, radiology reports, to name a few. There are a number of core
NLP tasks and machine learning models behind NLP applications.

Deep learning has recently brought a paradigm shift from traditional task-specific feature
engineering to end- to-end systems, and has obtained high performance across many
different NLP tasks and downstream applications. Tech companies like Google, Alibaba,
Apple, Amazon, Facebook, Tencent, and Microsoft are now actively working on deep
learning methods to improve their products. For example, Google replaced their traditional
statistical machine translation and speech recognition systems with systems based on deep
learning. The course provides a thorough introduction to cutting-edge methods in deep
learning applied to NLP. On the model side, we will cover word representations, window-
based neural networks for NLP, recurrent neural networks for NLP, long-short-term-memory
models, recursive neural networks for parsing, convolutional neural networks, as well as
more advanced topics like encoder-decoder models and attention mechanisms.

AI6128 Urban Computing

Urban computing is an emerging subject that aims to deal with the challenges in the rapid
urbanization progress such as pollution, energy consumption, and traffic congestion, etc.
Based on the Internet of Things (IoT) infrastructure and artificial intelligence, urban
computing acquires, integrates, and analyzes big and heterogeneous data generated by a
diversity of sources in urban spaces, such as sensors, devices, vehicles, buildings, and
human, to tackle the major urban challenges.

This course will introduce 1) the concept and applications of urban computing, 2) urban
sensing technologies including the acquisition of spatio-temporal urban data, 3) cloud
computing infrastructure to support urban computing applications, 4) urban data
management including spatial data management and spatio-temporal data management, 5)
ubran data mining including spatial data mining and spatio-temporal data mining, 6) urban
data learning including spatio-temporal prediction and reinforcement learning and 7)
advanced topics such as edge computing and cross-domain urban data fusion.

AI6129 AI Master Project

Course Coordinator: Asst Prof Yu Han [email protected]

This project, to be performed over two consecutive semesters, provides students an
opportunity to work with faculty members in SCSE and to learn state-of-the-art AI
techniques for a particular problem. Scope of project may be based on staff research
interest or industrial collaboration. Students are expected to document their work and
report their findings in formal reports, and give oral presentations together with
demonstration (if any) as the conclusion of their projects. The master project of AI is
a one year project with 6 AUs.
The master project of AI is a one-year project with 6 AUs.

Interested students are to register for AI6129 during the Course Registration Period
first, project is chosen in consultation with a supervisor (via the on-line MSAI project
application available via GSLink) during the first two weeks of the semester in which
the course will be taken. The project topic requires agreement by the proposed
supervisor.

Project Timeline
- Faculty Members propose Projects
- Students choose AI Master Projects
- Students work on the Projects
- Students submit the AI Master Project Report
- Supervisor and Examiner Evaluation
- Finalization of Marks and Marks Submission