Doctor of Philosophy in Machine Learning

Mode
Full-time

Credits
59

Location
On-campus

Overview

The scientific study of algorithms and statistical models that computer systems use to effectively perform a specific task without using explicit instructions, relying on patterns and inference instead. These algorithms are based on mathematical models learned automatically from data, thus allowing machines to intelligently interpret and analyze input data to derive useful knowledge and arrive at important conclusions. Machine learning is heavily used for enterprise applications (e.g., business intelligence and analytics), effective web search, robotics, smart cities and understanding of the human genome.

Program learning outcomes

Upon completion of the program requirements, the graduate will be able to:

  • Obtain rigorous mathematical background and advanced reasoning capabilities to express comprehensive and deep understanding of the pipelines at the frontier of machine learning: data, models, algorithmic principles and empirics.
  • Master a range of skills and techniques in data-preprocessing, exploration, and visualization of data-statistics as well as complex algorithmic outcomes.
  • Have a critical awareness of the capabilities and limitations of the different forms of learning algorithms and the ability to critically analyze, evaluate, and improve the performance of the learning algorithms.
  • Grow expert problem-solving skills through independently applying the principles and methods learned in the program to various complex real-world problems.
  • Develop a deep understanding of statistical properties and performance guarantees, including convergence rates (in theory and practice) for different learning algorithms.
  • Become an expert in using and deploying machine learning-relevant programming tools for a variety of machine learning problems.
  • Grow proficiency in identifying the limitations of existing machine learning algorithms and the ability to conceptualize, design, and implement an innovative solution for a variety of highly complex problems to advance the state-of-the-art in machine learning.
  • Able to initiate, manage, and complete research manuscripts that demonstrates expert self-evaluation and advanced skills in communicating highly complex ideas related to machine learning.
  • Obtain highly sophisticated skills in initiating, managing, and completing multiple project reports and critiques on a variety of machine learning methods, that demonstrates expert understanding, self-evaluation, and advanced skills in communicating highly complex ideas.

Completion requirements

The minimum degree requirements for the Doctor of Philosophy in Machine Learning is 59 credits, distributed as follows:

Core Courses Number of Courses Credit Hours
Core 4 15 Credit Hours
Research Thesis 1 36 Credit Hours
Elective Courses 2 8 Credit Hours

Core courses

Ph.D. in Machine Learning is primarily a research-based degree. The purpose of coursework is to equip students with the right skillset, so they can successfully accomplish their research project (thesis). Students are required to take COM701 as a mandatory course. They can select three core courses from a concentration pool of six in the list provided below:

Code Course Title Credit Hours
COM701 Research Communication and Dissemination

In this course, students will learn how to effectively communicate and disseminate their research findings, both orally and in written form, to the larger community. In addition to acquiring hard communication skills, students will also be familiarized with how these skills fit into a broader context, learning, for instance, the importance of peer review, how to select a journal or conference for publication, how to measure impact factor, how to gauge and adjust to different audiences, the various ethical issues that can arise, etc.

3
ML701 Machine Learning

This course provides a comprehensive introduction to Machine Learning. It builds upon fundamental concepts in Mathematics, specifically probability and statistics, linear algebra, and calculus. Students will learn about supervised and unsupervised learning, various learning algorithms, and basics of learning theory, graphical models, and reinforcement learning

4
ML702 Advanced Machine Learning

This course focuses on recent advances in machine learning and on developing skills for performing research to advance the state of the art in machine learning. Students will learn concepts in kernel methods, statistical complexity, statistical decision theory, computational complexity of learning algorithms, and reinforcement learning. This course builds upon concepts from Machine Learning (ML701) and assumes familiarity with fundamental concepts in machine learning, optimization, and statistics.

4
ML703 Probabilistic and Statistical Inference

Probabilistic and statistical inference is the process of drawing useful conclusions about data populations or scientific truths from uncertain and noisy data. This course will cover different modes of performing inference including statistical modelling, data-oriented strategies, and explicit use of design and randomization in analyses. Furthermore, it will provide an in-depth treatment of the broad theories (frequentists, Bayesian, likelihood) and numerous practical complexities (missing data, observed and unobserved confounding, biases) for performing inference. This course presents the fundamentals of statistical and probabilistic inference and shows how these fundamental concepts are applied in practice.

4
ML704 Machine Learning Paradigms

This course focuses on machine learning and on developing skills for performing research to the state of the art in machine learning. This course builds upon concepts from ML 701 and assumes familiarity with fundamental concepts in optimization, and statistics. Students will learn about methods in supervised, unsupervised learning, semi-supervised learning, transfer learning, multi-task learning, online learning, active learning, meta learning, and variational inference. The course will discuss variants of learning algorithms in various learning paradigms mentioned above.

4
ML705 Topics in Advanced Machine Learning

This course focuses on recent advances in machine learning and on developing skills for performing research to advance the state of the art in machine learning. This course builds upon concepts from ML701 and ML702 and additionally assumes familiarity with fundamental concepts in optimization, and math. The course covers advanced topics in statistical machine learning, unsupervised learning, high-dimensional statistics, and reinforcement learning. Students will be engaged through coursework, assignments, and projects.

4
ML706 Advanced Probabilistic and Statistical Inference

The study of probabilistic and statistical inference deals with the process of drawing useful conclusions about data populations or scientific truths from uncertain and noisy data. This course will cover some highly specialized topics related to statistical inference and their application to real-world problems. The main topics covered in this course are latent variable learning, kernel methods and approximate probabilistic inference strategies. This course will provide an in-depth treatment to various learning techniques (likelihood, Bayesian and max-margin) and numerous practical complexities (missing data, observed and unobserved confounding, biases) for performing inference.

4

Elective courses

Students will select a minimum of two elective courses, with a total of eight (or more) credit hours (CH) from a list of available elective courses based on interest, proposed research thesis, and career perspectives, in consultation with their supervisory panel. The elective courses available for the Doctor of Philosophy in Machine Learning are listed in the table below:

Code Course Title Credit Hours
MTH701 Mathematical Foundations for Artificial Intelligence

This course provides a comprehensive mathematical foundation for artificial intelligence. It builds upon fundamental concepts in linear algebra, probability theory, and basic statistics and overviews basics and advanced topics that are frequently encountered in AI applications. The students will learn the basic mathematical concepts for main AI systems, as well as realistic applications in AI of mathematical tools.

4
MTH702 Optimization

This course provides a graduate-level introduction to the principles and methods of optimization, with a thorough grounding in the mathematical formulation of optimization problems. The course covers fundamentals of convex functions and sets, 1st order and 2nd order optimization methods, problems with equality and/or inequality constraints, and other advanced problems.

4
AI701 Artificial Intelligence

This course provides the students a comprehensive introduction to modern artificial intelligence (AI), and some of its representative applications. The students will be familiarized with both the historical and recent AI techniques that have proven successful in building practical systems.

4
AI702 Deep Learning

This course provides a comprehensive overview of different concepts and methods related to deep learning. Students will first learn the foundations of deep learning, after which they will be introduced to a series of deep models: convolutional neural networks, autoencoders, recurrent neural network, and deep generative models. Students will work on case studies of deep learning in different fields such as computer vision, medical imaging, natural language processing, etc.

4
DS701 Data Mining

This course is an introductory course on data mining, which is the process of discovering patterns in large data sets involving methods at the intersection of machine learning, statistics, and database systems.

4
DS702 Big Data Processing

This course is an introductory course on big data processing, which is the process of analyzing and utilizing big data. The course involves methods at the intersection of parallel computing, machine learning, statistics, database systems, etc.

4
HC701 Medical Imaging: Physics and Analysis

This course provides a graduate-level introduction to the principles and methods of Medical Imaging, with thorough grounding in the physics of the imaging problems. This course covers the fundamentals of X-ray, CT, MRI, Ultrasound, and PET, imaging. In addition, the course provides an overview of 3D geometry of medical images and a few classical problems in medical images analysis including classification, segmentation, registration, quantification, reconstruction and radiomics.

4
CV701 Human and Computer Vision

This course provides a comprehensive introduction to the basics of human visual system and color perception, image acquisition and processing, linear and nonlinear image filtering, image features description and extraction, classification, and segmentation strategies. Moreover, students will be introduced to quality assessment methodologies for computer vision and image processing algorithms.

4
CV702 Geometry for Computer Vision

The course provides a comprehensive introduction to the concepts, principles and methods of geometry-aware computer vision which helps in describing the shape and structure of the world. In particular, the objective of the course is to introduce the formal tools and techniques that are necessary for estimating depth, motion, disparity, volume, pose and shapes in 3D scenes.

4
CV703 Visual Object Recognition and Detection

This course provides a comprehensive overview of different concepts and methods related to visual object recognition and detection. In particular, the students will learn a large family of successful and recent state-ofthe-art architectures of deep neural networks to solve the tasks of visual recognition, detection, and tracking.

4
CV704 Advanced Computer Vision

This course provides focused coverage of the following special topics: 1) image restoration and enhancement, 2) hand-crafted features, and 3) visual object tracking. The students will develop skills to critique the state-ofthe-art works on the aforementioned problems. Moreover, students will be required to implement papers with the aims of, (1) reproducing results reported in the papers and (2) improving performance of the published works. This course builds upon concepts from Human and Computer Vision (course code: CV701) and assumes familiarity with fundamental concepts in image processing

4
CV705 Advanced 3D Computer Vision

The course exercises an in-depth coverage of special topics in 3D computer vision. The students will be able to critique the state-of-the-art methods on 3D reconstruction, 3D visual scene understanding and multi-view stereo. In addition, students will have to implement papers to accomplish the following goals: (1) reproduce results reported in the papers, and (2) improve the performance of published peer-reviewed works. This course builds upon concepts from Human and Computer Vision (CV701), Geometry for Computer Vision (CV702) and assumes that the students are familiar with the basic concepts of machine learning and optimization.

4
CV706 Neural Networks for Object Recognition and Detection

This course provides focused coverage of special topics on object recognition and detection. The students will develop skills to critique the state-of-the-art works on visual object recognition and detection. Moreover, students will be required to implement papers with the following aims: (1) reproduce results reported in the seminal research papers, and (2) improve the performance of the published works. This course builds upon concepts from Human and Computer Vision (CV701), Visual Object Recognition and Detection (CV702) and assumes familiarity with fundamental concepts in machine learning and optimization.

4
NLP701 Natural Language Processing

This course provides a comprehensive introduction to Natural Language Processing. It builds upon fundamental concepts in Mathematics, specifically probability and statistics, linear algebra, and calculus, and assumes familiarity with programming.

4
NLP702 Advanced Natural Language Processing

This course provides a comprehensive introduction to Natural Language Processing. It builds upon fundamental concepts in Natural Language Processing and assumes familiarization with Mathematical concepts and programming.

4
NLP703 Speech Processing

This course provides a comprehensive introduction to Speech Processing. It builds upon fundamental concepts in Speech Processing and assumes familiarization with Mathematical and Signal Processing concepts.

4
NLP704 Deep Learning for Language Processing

This course focuses on recent advances in Natural Language Processing and on developing skills for performing research to advance the state of the art in Natural Language Processing. This course builds upon concepts from Natural Language Processing (NLP701) and assumes familiarity with fundamental concepts in Word Embedding, Information Extraction and Machine Translation.

4
NLP705 Topics in Advanced Natural Language Processing

This course focuses on recent advances in Natural Language Processing and on developing skills for performing research to advance the state of the art in Natural Language Processing. This course builds upon concepts from Natural Language Processing (course code: NLP 701) and assumes familiarity with fundamental concepts in question answering, text summarization and opinion mining.

4
NLP706 Advanced Speech Processing

This course focuses on developing skills for performing research to advance the state of the art in Speech Processing. This course builds upon concepts from Basic Speech Processing (NLP 703) and assumes familiarity with fundamental concepts in Speech Recognition, Speech Synthesis and Speaker Identification.

4

Research thesis

Ph.D. thesis exposes students to cutting-edge and unsolved research problems in the field of Machine Learning, where they are required to propose new solutions and significantly contribute towards the body of knowledge. Students pursue an independent research study, under the guidance of a supervisory panel, for a period of three to four years.

Code Course Title Credit Hours
ML799 Ph.D. Research Thesis

Ph.D. thesis research exposes students to cutting-edge and unsolved research problems, where they are required to propose new solutions and significantly contribute towards the body of knowledge. Students pursue an independent research study, under the guidance of a supervisory panel, for a period of three to four years. Ph.D. thesis research helps train graduates to become leaders in their chosen area of research through partly supervised study, eventually transforming them into researchers who can work independently or interdependently to carry out cutting-edge research.

36

Admission criteria

Bachelor’s degree in a STEM field such as computer science, electrical engineering, computer engineering, mathematics, physics and other relevant science and engineering majors, from a university accredited or recognized by the UAE Ministry of Education (MoE) which demonstrates academic distinction in a discipline appropriate for the doctoral degree. Students should have a minimum CGPA of 3.5 (on a 4.0 scale) or equivalent.

OR

Bachelor’s and master’s degrees in STEM fields such ascomputer science, electrical engineering, computer engineering, mathematics, physics and other relevant science and engineering majors, from a university accredited or recognized by the UAE Ministry of Education (MoE). Students should have a minimum CGPA of 3.2 (on a 4.0 scale) or equivalent.

Applicants must provide their completed degree certificates and transcripts (in English) when submitting their application. Senior-level students can apply initially with a copy of their transcript and expected graduation letter and upon admission must submit the official completed degree certificate and transcript. A degree attestation (for degrees from the UAE) or an equivalency certificate (for degrees acquired outside the UAE) should also be furnished within their first semester at the university.

Each applicant must show proof of English language ability by providing valid certificate copies of either of the following:

  • TOEFL iBT with a minimum total score of 90
  • IELTS Academic with a minimum overall score of 6.5
  • EmSAT English with a minimum score of 1550

TOEFL iBT and IELTS academic certificates are valid for two (2) years from the date of the exam while EmSAT results are valid for eighteen (18) months. Only standard versions (i.e. conducted at physical test centers) of the accepted English language proficiency exams will be considered.

Waiver requests from eligible applicants who are citizens (by passport or nationality) of UK, USA, Australia, and New Zealand who completed their studies from K-12 until bachelor’s degree and master’s degree (if applicable) from those same countries will be processed. They need to submit notarized copies of their documents during the application stage and attested documents upon admission. Waiver decisions will be given within seven days after receiving all requirements.

A valid and standard (i.e. exam was conducted at a physical test center) Graduate Record Examination (GRE) general test certificate with the following minimum scores is mandatory for applicants applying with only a bachelor’s degree:

  • 150 in verbal reasoning
  • 150 in quantitative reasoning
  • 3.0 in analytical writing

In an 800-word essay, please explain why you would like to pursue a graduate degree at MBZUAI and include the following information:

  • Motivation for applying to the university
  • Personal and academic background and how it makes you suitable for the program you are applying for
  • Stand-out achievements, e.g. awards, distinction, etc
  • Goals as a prospective student
  • Preferred career path and plans after graduation
  • Any other details that will support the application

The research statement is a document summarizing the potential research project an applicant is interested in working on and clearly justify the research gap which the applicant would like to fill in during the course of his/her study. It must be presented in the context of currently existing literature and provide an overview of how the applicant aims to investigate the underlying research project as well as predict the expected outcomes. It should mention the relevance and suitability of the applicant’s background and experience to the project and highlight the project’s scientific and commercial significance. The research statement should include the following details:

  • Title
  • Problem definition
  • Literature review
  • Proposed research/methods/solution (optional)
  • Study timeline (a table, figure or a small paragraph presenting your plans for the four years in the Ph.D. program)
  • List of references

Applicants are expected to write the research statement independently. MBZUAI faculty will NOT help write it for the purpose of the application. The MBZUAI selection team will review the submitted document and use it as one of the measures to gauge and assess applicants’ skills.

Applicants will be required to nominate referees who can recommend their application. Ph.D. applicants should have a minimum of three (3) referees wherein two were previous course instructors or faculty/research advisors and the other a previous work supervisor.

To avoid issues and delays in the provision of the recommendation, applicants have to inform their referees of their nomination beforehand and provide the latter's accurate information in the online application portal. Automated notifications will be sent out to the referees upon application submission.

Selected applicants will be invited to participate in an entry exam that will include questions related to the following topics:

  • Math: Basic math questions related to calculus, probability theory, linear algebra and optimization.

Ph.D. applicants are recommended to read about these topics especially on how they are related to machine learning. Online research is encouraged, for instance, searching for “linear algebra for machine learning” on YouTube for useful video lectures. In addition, they can find courses on sites like Coursera, Udemy, and many others. Applicants are encouraged to review as many resources as possible. The following website is recommended for mathematics for machine learning: https://mml-book.github.io

  • Machine learning:

Ph.D. applicants are expected to have basic understanding of different machine learning algorithms and concepts such as linear regression, decision trees, loss functions, support vector machines, classification, regression, clustering, convolutional neural networks, etc is important. They don’t have to master these concepts, but they need to have basic knowledge. There are many online courses and tutorials to get Ph.D. applicants prepared. Useful video lectures are available on YouTube, Coursera, Udemy, and many others. Applicants are encouraged to review as many resources as possible. The following website is recommended for deep learning: https://www.deeplearningbook.org/

  • Programming

Strong programming background is important for Ph.D. applicants. They will be asked programming questions. Most questions are in Python but the specific language is not a problem since the questions are algorithmic rather than language-specific. Therefore, it is important to read about different data structures such as Arrays, Stacks, Queues, etc. It is also important to read about different programming algorithms such as sorting and searching algorithms, and complexity. Ph.D. applicants will be asked questions which require finding the output of a piece of code, finding the problem/error in a short code, and finding the code which performs a specific task.

The exam instructions are available here.
A technical admission interview with MBZUAI faculty will follow.

Study plan

A typical study plan is as follows:

Semester 1

COM701 Research Communication and Dissemination
ML701 Machine Learning*
+ 1 Elective

Semester 2

ML702 Advanced Machine Learning*
ML703 Probabilistic and Statistical Inference*
+ 1 Elective

Semesters 3-8

ML799 Master’s Research Thesis
* In consultation with the student’s supervisor, the following advanced courses may be substituted if classes are available:

ML704 Machine Learning Paradigms
ML705 Topics in Advanced Machine Learning
ML706 Advanced Probabilistic and Statistical Inference

Study Plan

DOWNLOAD

Career prospects

AI is permeating every industry. At recent employer engagement events at MBZUAI, there has been representation from multiples sectors including (but not limited to):

  • Aviation, consultancy, education, energy, finance, government entities, healthcare, media, oil and gas, security and defense, research institutes, retail, telecommunications, transportation and logistics, and startups.

Recent job opportunities advertised via the MBZUAI Student Careers Portal include (but not limited to):

  • AI solution architect, AI solution engineer, algorithmic engineer, data analyst, data engineer, data scientist, data strategy consultant, full stack software engineer, full stack web developer, predictive analytics researcher, and senior data scientist - consultant.

Other career opportunities could include (but not limited to):

  • Applied scientist, analytics engineer, augmented/virtual reality, autonomous cars, biometrics and forensics, chief data officer, data platform leadership, data journalist, data and AI technical sales specialist, growth analytics / engineers, manager: AI and cloud services planning, machine learning engineers, product manager: AI and data analytics, product data scientist, product analyst, remote sensing, research assistants, security and surveillance, senior software engineer, and VP data.

Meet the faculty

...

Professor Eric Xing

President
and Professor of machine learning

READ BIO
...

Professor Le Song

Department Chair and Professor,
Machine Learning

READ BIO
...

Professor Mohsen Guizani

Professor,
Machine Learning

READ BIO
...

Professor Mérouane Debbah

Professor,
Machine Learning

READ BIO
...

Professor Najwa Aaraj

Professor
Machine Learning

READ BIO
...

Dr. Martin Takac

Associate Professor,
Machine Learning

READ BIO
...

Dr. Kun Zhang

Assistant Professor,
Machine Learning

READ BIO
...

Dr. Huan Xiong

Assistant Professor,
Machine Learning

READ BIO
...

Dr. Bin Gu

Assistant Professor,
Machine Learning,

READ BIO
...

Dr. Qirong Ho

Assistant Professor,
Machine Learning,

READ BIO
...

Dr. Shangsong Liang

Assistant Professor,
Machine Learning,

READ BIO
...

Dr. Zhiqiang Xu

Assistant Professor,
Machine Learning,

READ BIO
...

Dr. Pengtao Xie

Assistant Professor,
Machine Learning,

READ BIO

Disclaimer: Subject to change.