The MINDS iPhD is a four-year integrated doctoral programme combining a structured first year of taught training with three years of supervised research.

It was designed around a specific insight: that the most important problems at the intersection of machine intelligence and nanoelectronics require researchers who are genuinely fluent in both fields; not specialists who have crossed briefly into adjacent territory.

The first year builds that fluency deliberately, through a combination of core MINDS modules and a tailored selection of advanced courses drawn from computer science and electronics. The second to fourth years are the research itself individual doctoral projects, each associated with one or more of the CDT's three research themes, each supervised by an interdisciplinary team that typically includes an industry advisor.

Intake was approximately 12 researchers per cohort, across five cohorts from 2019.

Year 1: Training and foundations

The first year of the MINDS programme was structured to give every researcher a common grounding across machine intelligence and nanoelectronics; regardless of whether their undergraduate background was primarily in computer science, electronics or engineering. Two core MINDS modules ran across the year, with a tailored selection of advanced modules chosen in discussion with supervisors to match each researcher's background and research direction.

Semester 1

Two core MINDS modules formed the spine of the first semester, with one ELEC (electronics) and one COMP (computer science) optional module selected to match each researcher's prior knowledge and intended research direction.

MIND6001 Impact and Communication Strategies
MIND6002 Technologies, Modules and Algorithms
COMP6203 Intelligent Agents
COMP6246 Machine Learning Technologies
COMP6245 Foundations of Machine Learning
ELEC6254 Foundations of Embedded IoT Systems
ELEC6237 Secure Hardware and Enbedded Devices
ELEC3207 Nanoelectronic Devices

Semester 2

The Interdisciplinary Team Project (MIND6003) was the centrepiece of Semester 2 a collaborative module that brought researchers from across the cohort together to tackle a shared challenge. Two additional COMP or ELEC modules were selected to deepen knowledge in the areas most relevant to each researcher's emerging project.

MIND6003 Interdisciplinary Team Project
COMP6207 Algorithmic Game Theory
COMP6208 Advanced Machine Learning
COMP6257 Bayesian, Active & Reinforcement Learning
COMP6258 Differentiable Programming and Deep Learning
ELEC6207 Quantum Devices and Technology
ELEC6214 Advanced Wireless Communications Networks and Systems
ELEC6232 Analogue and Mixed Signal CMOS Design
ELEC6234 Embedded Processors

Summer

The summer of Year 1 marked the transition from taught training to doctoral research. Researchers began working directly with their supervisory team on MIND6004 the Feasibility Study, which scoped the individual research project that would carry through Years 2 to 4.

Over the summer in Year 1, you will start working with your research project supervisors on MIND6004 (Feasibility Study).

Years 2 - 4: Doctoral research

Years 2 to 4 of the MINDS programme were the core research phase. Building on the feasibility study completed at the end of Year 1, each researcher worked with a dedicated supervisory team typically including academics from across disciplines and an industry advisor to develop and complete an original doctoral project.

All MINDS projects were associated with one or more of the CDT's three research themes and took a cross-disciplinary approach as a matter of programme design, not just aspiration. Industry and government partners had active input into project supervision, and researchers had access to the full range of MINDS facilities and networks throughout.

The emphasis throughout was on research that was genuinely disruptive; work that crossed disciplinary boundaries while remaining aware of its potential implications for society, security and safety.

Research Themes

MINDS research was organised across three themes, each addressing a different aspect of the same underlying challenge: how to make AI work intelligently, efficiently and reliably in physical devices and distributed systems. Most projects drew on more than one theme, reflecting the CDT's emphasis on cross-disciplinary approaches.

Agent Based Adaptive Systems (ABAS)

As AI moves into edge devices and distributed networks, the question of how large numbers of autonomous systems coordinate; without centralised control, and without compromising security or trust becomes critical. ABAS research in MINDS investigated the algorithms and architectures that make this possible.

Research questions included: how can machine learning be effectively decentralised? How do autonomous agents coordinate securely and build trust with human operators? How do novel device capabilities open new possibilities for agent-based AI? Specific challenges addressed included federated learning, game-theoretic models for optimising collective behaviour, automated responses to cyber-attacks in distributed systems, and the design of AI that respects local autonomy while guaranteeing system-level performance.

Embedded Artificial Intelligence (EAI)

Current AI models particularly deep neural networks have computational demands that make them difficult or impossible to run on low-power edge devices. The EAI theme addressed this directly: how do you design AI that is genuinely efficient, secure and reliable when running on constrained hardware?

Research focused on energy and memory-efficient AI and ML algorithms, hardware-software co-design, and techniques for moving AI processing to the network edge without sacrificing performance or security. Work in this theme connected directly to the CDT's nanoelectronics and device physics capabilities, developing AI approaches that were shaped by the physical realities of the devices they would run on.

Nanoelectronic Technologies for AI (NTAI)

AI processing makes fundamentally different demands on hardware than conventional computing favouring architectures built around multiply-accumulate operations, adjustable memory and co-located memory and computation. Standard semiconductor technologies are not optimised for these tasks, and that gap affects performance, efficiency and scale. NTAI research investigated alternatives at the materials and device level.

Projects in this theme focused on the invention, fabrication, characterisation and optimisation of emerging device technologies that exploit nanomolecular effects, including resistive switching, where memory is stored in a highly confined volume using minimal energy to shift just a few atoms between states. Work developed understanding of the electrochemistry underpinning these technologies and used that understanding to build new fabrication processes, characterisation techniques and device architectures suited to AI workloads.

Task-Optimised Devices and Systems (TODS)

Rather than adapting general-purpose hardware to run AI, TODS research asked a different question: what does hardware look like if it is designed specifically for machine intelligence tasks from the start? This theme sat at the intersection of device physics, materials science and AI systems design.

Research spanned new device architectures, nanofabrication approaches, and system-level designs built around AI performance characteristics. It drew extensively on the CDT's access to the Zepler Institute Nanofabrication Cleanroom, enabling researchers to move from theoretical device concepts to physical prototypes. Applications ranged from neuromorphic computing and low-power inference hardware to novel sensor systems and next-generation semiconductor devices.

Training and Support

Training and Support

Research at the interface between AI and electronics requires thinking differently about the interdependencies between the two disciplines their underpinning mathematics, statistics, algorithmic techniques and device physics. The MINDS CDT was structured to build that breadth deliberately, drawing on expertise across both fields.

Cross-Disciplinary Expertise

MINDS researchers gained broad awareness across AI and electronics through a combination of core and tailored modules, selected to match each researcher's academic background and research direction. All researchers followed a combination of computer science and electronics modules covering areas including deep learning, online and reinforcement learning, intelligent agents, nanoelectronic devices, embedded processors, and analogue and mixed signal systems design. This cross-disciplinary foundation was put into practice through a team project addressing a challenge set by one of the CDT's industry or government partners, with outcomes demonstrated at the annual showcase.

Innovation Camp

Each year, MINDS researchers attended a week-long, student-led innovation camp. Activities ranged from enhanced technical training and research commercialisation workshops run with SetSquared, to the development of outreach activities and, inevitably, social ones.

STARs Programme

The STARs programme gave MINDS researchers a structured framework for short-term, focused collaborative research with peers across the CDT. Researchers could connect with others working in related areas to explore new ideas, exchange best practice, learn new techniques and tools, or gain perspective across disciplines. Industry partners were included in the network.

Outreach & Public Engagement

All MINDS researchers took part in outreach and public engagement activity, communicating the societal value of AI and electronics research and inspiring others toward the field. Through the CDT's Ambassadors' Programme, several researchers received enhanced training in outreach leadership.