Is this course right for me?

Target Audience: Academics/Researchers/KE Staff

This seventh workshop is part of a series of workshops exploring ML/AI opportunities for collaboration with university colleagues, run between the faculties of Science, Engineering and the Business School.  Prior knowledge of the topic is not required.

The Programme:

 1:05 – 1:10   Welcome Professor Sergey Kitaev

 1:10 – 1:30 Industrial AI Cluster at Strathclyde

Professor Martin Halvey (Head of Computer and Information Sciences; Industrial AI Cluster Lead)

 1:30 – 1:45 Physics-Informed Machine Learning Approaches for Material-to-Product Prediction of Tablet Properties

Dr Mohammad Salehian (CMAC)

 1:45 – 2:00 Human-Centered Interactive Intelligence

Dr Yingying Zhao (Computer and Information Sciences)

 2:00 – 2:15 Deep Learning Image Segmentation for Pharmaceutical Particle Characterisation

Dr Christopher Boyle (CMAC)

 2:15 – 2:30 Computational Modelling of Robots and Molecules

Dr Didier Devaurs (Computer and Information Sciences)

 2:30 – 2:35 ARCHIE-WeSt: A Platform for AI & Machine Learning at Strathclyde

Dr Richard Martin (Manager, ARCHIE-WeSt)

 2:35 – 2:55 Round up, networking and discussion

Should you have any questions about the workshop, please contact Associate Dean Research Professor Sergey Kitaev or Dr Graeme West or Professor Matthew Revie.

Abstracts:

Speaker: Professor Martin Halvey (Head of Computer and Information Sciences; Industrial AI Cluster Lead)

Title: Industrial AI Cluster at Strathclyde

Abstract: All organisations wish to accelerate their digitalisation journey, unlock insights from their data and take commercial advantage of the advances in Artificial Intelligence (AI) and analytics to grow business value, or achieve strategic goals. For industrial sectors, AI and informatics technologies present particular opportunities. These include increased operational and process efficiencies; extension of equipment lifetime; cost-effective asset management; and enhanced safety. Achieving these business benefits can incur challenges with legacy assets or equipment; multiple data sources; complex operational environments; and the appropriate in-house technical skills. This brief talk introduces the Industrial AI Cluster at Strathclyde. This cluster brings together expertise from across the University of Strathclyde, and our partners, to solve industrial challenges and find technology and AI solutions which will add value to organisations across a range of industrial sectors.

---------------------------------

Speaker: Dr Mohammad Salehian (CMAC)

Title: Physics-Informed Machine Learning Approaches for Material-to-Product Prediction of Tablet Properties

Abstract: Pharmaceutical tablet formulation and process development, traditionally complex and multi-dimensional decision-making process, necessitates extensive experimentation and resources, often resulting in suboptimal solutions. To address these challenges, there has been an increasing use of hybrid modelling (data-driven and mechanistic) to accelerate and increase resource efficiency of the development of drug products. The limitations of currently developed methodologies are: 1) The absence of a comprehensive system of models that can predict blend properties (i.e. mixture models) and final product attributes (i.e. process models) based on raw component properties, process conditions, and formulation; 2) A lack of large-scale optimisation framework to validate models and achieve a product with the desired critical quality attributes. We propose a hybrid modelling framework based on physics-guided data-driven models for the development of directly compressed tablets in pharmaceutical manufacturing to achieve the optimal drug product quality while minimising development time and cost.

---------------------------------

Speaker: Dr Yingying Zhao (Computer and Information Sciences)

Title: Human-Centered Interactive Intelligence

Abstract: Understanding human intelligence and creating interactive technologies that center on human needs are not just academic pursuits; they form the foundation of a future where technology truly serves people. In this talk, I will present three interconnected projects advancing the field of human-centered interactive intelligence: 1) investigating canine causal attention by integrating animal vision and computer vision insights; 2) decoding how people read by merging visual processing with textual semantics; and 3) exploring human behaviors and personalities through the fusion of computer vision, large language models (LLMs), human vision, and speech recognition. These projects go beyond theoretical exploration, delivering practical applications that improve user experiences. From reading assistance tools to LLM-powered companions for emotional support, we strive to bridge cutting-edge science with real-world impact. Rooted in interdisciplinary collaboration, this journey underscores the role of technology by placing human needs at its core. I will conclude by outlining potential future directions and our ongoing efforts to expand the impact of human-centered interactive intelligence.

---------------------------------

Speaker: Dr Christopher Boyle (CMAC)

Title: Deep Learning Image Segmentation for Pharmaceutical Particle Characterisation

Abstract: A key process in pharmaceutical manufacturing is crystallisation wherein an Active Pharmaceutical Ingredient (API) is isolated and purified into solid crystals. The shape and size of these crystals defines in part how the medicine performs in the patient. *In situ* microscopy is a promising method for measuring particle characteristics due to the wealth of information that can be gleaned from an image. Image analysis is made difficult by the relatively extreme changes in brightness and contrast through an experiment as well as the high density of crystals which may overlap on the image. In this work, we are developing methods for analysing images using deep learning models. There are few relevant datasets available for this task, so image annotation is a big part of the work. We are looking at techniques like transfer and contrastive learning to reduce the annotation burden.

---------------------------------

Speaker: Dr Didier Devaurs (Computer and Information Sciences)

Title: Computational Modelling of Robots and Molecules

Abstract: In the past 14 years, my research has focused on computationally modelling, simulating and analysing complex physical systems, both in robotics and structural biology. At the algorithmic level, simulating mobile or flexible systems (such as robots and molecules) requires exploring a high-dimensional space: the space of all possible states of the system. My work has involved developing efficient algorithms and heuristics to address this challenge. During my PhD, I developed novel extensions of sampling-based path planning algorithms by creating the concept of optimal path planning in a cost space. As a post-doctoral researcher, I then developed computational methods for the efficient conformational sampling of molecular systems, such as large proteins and protein-peptide complexes.
I now focus on biomedical AI to produce computational methods and tools involving state-of-the-art AI techniques informed by biomedical data, which will help clinicians create novel personalised medicine strategies. The innovation of my research lies in integrating experimental data within cutting-edge AI techniques. One of my short-term goals is to develop computational tools to predict the clinical interpretation (as benign or pathogenic) of genetic mutations, using data produced by so-called deep mutational scanning experiments. My longer-term goals involve combining mutational and structural biological data to dissect mechanisms of action of mutations and perform computer-assisted protein engineering. For that, I will adapt state-of-the-art protein language models and generative AI methods whose explainability I will strive to increase through inputs from clinicians.

---------------------------------

Speaker: Dr Richard Martin (Manager, ARCHIE-WeSt)

Title: ARCHIE-WeSt: A Platform for AI & Machine Learning at Strathclyde

Abstract: ARCHIE-WeSt is a research computing platform based at 
Strathclyde which is available to all staff and collaborators, including 
industry partners. We will provide an overview of our hardware and 
software for supporting AI & ML based research.

Prerequisites