2-2.30pm: Alice Malivert
Title: Image-based AI diagnosis platform for early drought stress detection in plant leaves
Abstract: Drought stress, a lack of accessible water caused by reduced precipitation, salinity, wind and extreme temperatures, is the primary cause of crop loss worldwide and is expected to become more frequent and severe with climate change: currently, drought stress is responsible for over 34% of agricultural production loss in least developed and developing countries. To reduce the drought stress-related issues in agricultural production, we need to be able to detect the first signs of drought stress in plants. For that, researchers and farmers alike need a tool that would be fast, simple to use and cost-efficient while providing accurate quantitative measures. I propose to develop an AI assisted tool to detect early signs of drought stress in plant leaf pictures. Ultimately, this project will result in an open online platform to diagnose drought stress in new plant leaf pictures, as an affordable and accessible tool shared with the research and agricultural community.
2.30-3pm: Andreas Joergensen
Title: Efficient Bayesian Inference for Stochastic Agent-based Models
Abstract: Mathematical models can help researchers understand tumour growth and develop patient-specific treatments. For instance, one might draw on agent-based models that track individual cells or cell clusters to study cancer on a microscopic level. Agent-based models successfully recover the observed macroscopic growth patterns of tumours and provide insights into the underlying mechanisms. However, agent-based models are computationally expensive, which might render parameter inferences through Bayesian sampling schemes insurmountable. Moreover, the models are stochastic, i.e. the model predictions change if the simulations are repeated. Indeed, this complication arises for many biological systems and can become yet another stumbling block for inference algorithms. In the talk, I will discuss how machine learning can help us overcome computational constraints while still accounting for the intrinsic stochasticity of biological systems.
3.30-4pm: Fan Zhang
Title: Perception and Manipulation for Robot-Assisted Dressing
Abstract: Assistive robots have the potential to support people with disabilities in a variety of activities of daily living, such as dressing. This work involves using computer vision and machine learning algorithms to allow an assistive robot to understand the deformable cloth configurations for semantic segmentation and depth estimation, identify cloth physics, visual-tactile-based grasp/manipulate deformable objects, infer human user cognitive states and actions, learn personalized human motion model, and accordingly adapt its assistance to realize variable levels of autonomy for assistive robotics. We have tested our robot in a scenario that closely mimicked the Certified Nursing Assistant test used in US healthcare, in which a trainee nurse (in our case, a robot) has to go around the hospital bed and put an open-backed robe on a person with weak or paralyzed arms. Our work has been published on Science Robotics.
4-4.30pm: Jonathon Langford
Title: Understanding the Higgs boson at the Large Hadron Collider
Abstract: The Higgs boson resides at the centre of our current best theory for describing the elementary particles and their interactions. In this theory, known as the standard model (SM) of particle physics, the Higgs boson plays a key role in the symmetry breaking of the electroweak interaction, and the subsequent generation of mass for other elementary particles. This elusive particle was observed experimentally in 2012 by the ATLAS and CMS experiments at the Large Hadron collider, marking the completion of the particle content of SM theory. Since then, both experiments have worked tirelessly to further understand the properties of the Higgs boson and its interactions, to help further elucidate the nature of our universe. In this talk I will introduce the analyses that I have recently worked on, focussing on the statistical combination of Higgs boson measurements across decay channels, which was published in Nature in 2022. I will then focus on my current work, which involves applying a variety of Machine Learning methods to improve our sensitivity to measurements of Higgs boson properties and beyond.
Refreshments available between 15:00 – 15:30