ML Club Space Technology Special

ML Club
ML Club
Public group
Location image of event venue


We are honoured to have 3 great speakers for our ML Club Space Technology Special.

Callum Wilson – PhD student in the ICELAB at Strathclyde studying intelligent control techniques for on-board spacecraft control

Intelligent control combines theories and methods from artificial intelligence, operations research, and automatic controls. We use intelligent control techniques where a control system must operate without human intervention, or it is desired to improve the system’s performance during operation. This field is very well studied and there are a wide variety of existing intelligent control methods with numerous applications. Looking towards the future of intelligent control, why should we use intelligent control instead of conventional control methods, and what gaps are present for future research?

Audrey Berquand - PhD student at the Intelligent Computational Engineering (ICE) lab, University of Strathclyde within the Design Engineering Assistant project

Every space mission starts with a feasibility study. Since the mid 1950s, we have been accumulating knowledge on space mission design. Today, this enormous amount of data is available in an unstructured or a semi-structured format (e.g., reports written in natural language or engineering models developed with design softwares). The goal of the Design Engineering Assistant is to allow the experts involved in the early stages of space mission design, especially in the frame of concurrent engineering studies, to get a fast and reliable access to this data. Via the development of a space mission design ontology and an expert system based on, this project is bringing AI into the design process of space missions.

Siva - Big Data Analyst at Bird.i, a Geospatial company based in Glasgow.

Inception is a deep convolutional network that was developed by Google for the ILSVRC 2014. The Inception network stands on its own right for image classification, and can and has been used in object localisation networks as a backbone. The network also works very well as a feature extractor for training using regular ML algorithms such as SVM or Random Forests. The aim of the talk is to warm up to more technical talks in the year ahead, to discuss the network and design principles, and skim the heavy math.