Oxford Machine Learning Research Group

The Machine Learning Research Group (MLRG) at the University of Oxford is part of the Department of Engineering Science, specifically within Information Engineering. It plays a central role in Oxford’s Machine Learning and AI community. The group is closely linked with the Oxford-Man Institute of Quantitative Finance, and several faculty members hold joint affiliations with both institutes.

MLRG focuses on developing robust machine learning techniques and applying them to various fields such as science, engineering, and commerce. Their research covers Bayesian modelling, machine learning on graphs, reinforcement learning, neural networks, natural language processing, and optimisation. They aim to address complex real-world problems, including those in astronomy, finance, and disaster management.

The group is known for its contributions to several high-profile projects. These include ORCHID, which explores human-agent collectives, and Project HumBug, which uses AI to detect mosquitoes through low-cost smartphones to help fight malaria. Other notable projects focus on exoplanet detection, human-agent collectives, and intelligent systems for ecological monitoring.

The group’s faculty includes leading academics such as Professors Stephen Roberts, Michael Osborne, and Natalia Ares, among others. They are involved in cutting-edge research and are widely recognised in their fields. MLRG regularly publishes its findings in top conferences and journals like NeurIPS, ICML, and AISTATS.

MLRG is also involved in education and outreach. They offer courses in machine learning through Oxford’s graduate programmes, particularly focusing on probabilistic methods and their applications. The group runs seminars, workshops, and reading groups to foster knowledge-sharing within the academic community. Additionally, they provide software tools and resources for machine learning research, including various toolboxes for Bayesian methods.

The Machine Learning Research Group (MLRG) at the University of Oxford is based within the Department of Engineering Science, with strong links to the Oxford-Man Institute of Quantitative Finance. The group has a significant focus on the development of machine learning methods and their application to problems in science, engineering, and commerce.

MLRG was established as part of the University’s ongoing focus on machine learning research and its application across various fields. Machine learning and artificial intelligence (AI) have become important topics of study within the Department of Engineering Science, driven by the need for more robust and scalable solutions to real-world problems. Early research efforts focused on probabilistic modelling, Bayesian methods, and neural networks. The group’s initial contributions included developing tools for Bayesian inference and statistical learning, which have become widely used in various industries.

In the early 2010s, MLRG expanded its focus and took on significant research projects, particularly in Bayesian theory, graph-based machine learning, and reinforcement learning. During this period, the group also began developing models for practical applications in fields such as finance, astronomy, and signal processing.

One of the group’s most impactful partnerships began with the Oxford-Man Institute of Quantitative Finance, where machine learning methods were applied to financial problems such as predictive modelling and risk assessment. This collaboration strengthened the group’s research in financial technology (fintech) and led to numerous publications in top-tier journals and conferences.

From 2015 onwards, MLRG expanded its research to include applications in diverse fields such as autonomous systems, ecology, and mathematical zoology. One notable project, HumBug, used machine learning for mosquito detection via smartphones, contributing to public health efforts in controlling mosquito-borne diseases. Another key project focused on astrostatistics, where machine learning methods were used for exoplanet detection and transient analysis.

During this time, the group also began exploring human-agent collectives (HACs) through projects such as ORCHID, aiming to enhance collaboration between human agents and autonomous systems. The goal was to develop systems that allow humans and intelligent software agents to work together more effectively, particularly in disaster management and other dynamic, uncertain environments.

In recent years, MLRG’s research has increasingly focused on more complex areas such as probabilistic numerics, reinforcement learning from pixels, and neural network optimisation. The group has continued to make advances in autonomous intelligent systems, integrating multi-agent systems with Bayesian inference and model-predictive control to create robust models capable of handling real-world uncertainty. These models have been applied to fields such as robotics and distributed information networks.

MLRG has maintained a strong presence in the academic community by contributing to major conferences such as NeurIPS, ICML, and AISTATS, where their work has been recognised for its contributions to both theoretical and applied machine learning. The group has also remained committed to education, with programmes such as the Centre for Doctoral Training in Autonomous, Intelligent Machines and Systems (AIMS) and the Schmidt AI in Science Postdoctoral Fellowship programme providing advanced training to new researchers.

As of 2024, the Machine Learning Research Group remains a key player in the global machine learning research community. The group’s work spans numerous domains, from finance to ecology, and continues to address real-world challenges through machine learning and AI. Faculty members such as Professor Stephen Roberts and Professor Michael Osborne lead cutting-edge projects, while the group also fosters collaborations with industry partners such as Mind Foundry, a machine learning company co-founded by group members.

MLRG continues to publish extensively and participates in key research initiatives like Intelligent Earth, contributing to global efforts to advance AI and machine learning in both scientific and industrial applications. The group remains committed to solving real-world problems with innovative machine learning solutions, making substantial contributions to both the academic world and practical, applied machine learning technologies.

The Machine Learning Research Group (MLRG) at the University of Oxford offers a range of detailed courses as part of its educational programmes, aimed at both Master’s and PhD students. These courses focus on machine learning techniques, their applications, and the latest advancements in the field. The group provides courses as part of the MSc in Engineering Science and also contributes to the Centre for Doctoral Training (CDT) in Autonomous Intelligent Machines and Systems (AIMS).

MSc in Engineering Science

As part of the MSc in Engineering Science, MLRG offers courses that cover key machine learning topics. Students are introduced to both the theory and practical applications of machine learning methods. Key areas covered in this course include:

• Bayesian Modelling: This course introduces students to Bayesian methods, a key statistical approach used to model uncertainty in machine learning. Bayesian modelling is used in fields like finance, healthcare, and robotics, where making predictions based on uncertain data is crucial. Students learn how to use these models to make better decisions in complex environments.
• Reinforcement Learning: Reinforcement learning focuses on teaching systems to learn from their actions and make decisions based on feedback from their environment. It is particularly useful in developing AI systems that need to perform tasks autonomously, such as robots or self-driving cars. The course explores both the theory behind reinforcement learning and practical algorithms used in the field.
• Neural Networks: This course covers the basics and advanced aspects of neural networks, which are used to create models that mimic the way human brains process information. Neural networks are at the core of many modern machine-learning applications, including image recognition, natural language processing, and speech recognition. Students are taught how to design, train, and optimise neural networks.
• Natural Language Processing (NLP): NLP is the study of how machines can understand and process human language. This course covers the techniques used to enable computers to analyse text and speech, such as sentiment analysis, translation, and voice commands. NLP is widely used in technologies like virtual assistants and chatbots.
• Bayesian Optimisation: Bayesian optimisation is used to fine-tune machine learning models by finding the best set of parameters for a given problem. This is important in areas like robotics and control systems, where models need to be highly accurate. The course explores various techniques to optimise models efficiently.

Centre for Doctoral Training (CDT) in AIMS

For PhD students, MLRG is heavily involved in the CDT in Autonomous Intelligent Machines and Systems (AIMS). This programme provides specialised research training in the development of autonomous systems, combining machine learning techniques with real-world applications. Key areas of study include:

• Decision Theory: Decision theory focuses on making optimal decisions in uncertain environments. PhD students learn how to apply this theory to develop autonomous systems, such as drones, that can make decisions on their own while navigating through complex environments.
• Probabilistic Methods: This course teaches PhD students how to work with uncertain data and make predictions using advanced probabilistic methods. These skills are essential for solving real-world problems where data is incomplete or noisy, such as in environmental monitoring or financial forecasting.
• Multi-agent Systems: In this course, students study how multiple AI systems (agents) can interact and collaborate to solve complex problems. This is especially relevant for systems where many autonomous units need to work together, such as in disaster response or smart city infrastructure.

Seminars and Workshops

In addition to formal courses, MLRG organises regular seminars and workshops. These are designed to keep students and researchers up-to-date with the latest developments in machine learning. Experts from across the world are invited to present their work, and students are encouraged to engage in discussions and collaborations. Workshops focus on specific topics such as Bayesian inference, deep learning, and reinforcement learning.

Here are the global rankings for the Oxford Machine Learning Research Group:

• Ranked as one of the Top Artificial Intelligence Research Labs in the World alongside the Alan Turing Institute and J.P. Morgan AI.
• In the 2021 Research Excellence Framework (REF), 81% of the group's research was rated world-leading, with the rest rated internationally excellent.
• The University of Oxford is ranked 20 in Best Universities for Artificial Intelligence globally by U.S. News & World Report.
• Oxford is ranked 4 in Best Global Universities overall by U.S. News & World Report.

• Machine Learning Research Group - Information Engineering| University of Oxford
• Machine Learning Research Group| University of Oxford Department of Engineering Science
• research theme: Artificial Intelligence and Machine Learning| Oxford Department of Computer Science
• Oxford University Machine Learning Group| Imperial College London
• Machine Learning| Oxford Department of Computer Science
• Machine Learning and Economics Group, Oxford| GitHub Pages
• Computer Vision and Machine Learning| University of Oxford Department of Engineering Science
• Artificial Intelligence and Machine Learning : publications| Oxford Department of Computer Science
• Oxford University Machine Learning Summer School| DTU Research Database
• Machine Learning and Data Science | Mathematical Institute| University of Oxford