Data Bytes 1: AI & Automation

The 1st instalment of our data science mini-series showcasing current research projects carried out by the Data Scientist Development Programme at Leeds Institute for Data Analytics (LIDA).

This week’s features 10-minute lightning talks from two data scientists on using data analytics to AI & Automation.

Each talk will last 10mins + 10mins Q&A.

AGENDA

Talk 1 - Leveraging Data Analytics and Automation to Enhance Cyber Security Incident Investigation and Response (Hal Kolb)

Talk Summary: This project reviewed an extensive set of incident response data to investigate strategies to improve the speed, quality and efficiency of incident response and improve practices around data structuring and insight extraction. Following a thorough data analysis and preparation phase, this process identified potential in focusing effort on the classification of potential compromise indicators from Windows Event Logs. Through the consolidation of the disparate data into a robust, structured dataset, it was possible to apply and evaluate a series of anomaly detection techniques. A secondary benefit of this workstream resulted in the production of a proof-of-concept for automated extraction of cyber threat intelligence from natural language incident reports.

Speaker Biography: After completing their undergrad, driven by a commitment to driving positive change, Hal completed the Teach First training programme. Teaching was extremely rewarding however, Hal felt they could have a greater impact using their technical skills. Motivated by this goal, Hal pursued an MSc before joining the DSDP at LIDA.

Talk 2 - Interpretable AI for Pedestrian Intention Prediction: A Data-Driven Visualisation Approach (Piyush Mohan)

Talk Summary - Predicting pedestrian crossing intention is critical for safe autonomous vehicles, yet current AI models operate as black boxes, making them hard to evaluate. This project aims to develop an interactive visualisation dashboard to make Pedestrian Intention Prediction (PIP) models more interpretable and transparent. Using open datasets such as PIE, JAAD, and Waymo, the dashboard visualises how features like pedestrian pose, motion, and road context influence predictions. Building on MAVIS and Hi-Drive research, the work highlights model behaviour, biases, and failure cases to support safer AI-driven mobility. The approach contributes to public trust, road safety, and data-driven decision-making for transport systems.

Speaker Bio - Piyush is a Data Scientist at LIDA with a background in computer science and an MSc in Data Science and Analytics from the University of Leeds. He is currently working on visual analytics and explainable AI for pedestrian intention prediction models used in AV vehicles, applying data science to real-world transport challenges.

Also check out other Business events in Leeds, Workshops in Leeds.