for Industry

Building AI applications involves a large number of design decisions for developers. The number and intricacy of decisions grow with complexity, making even seasoned experts overwhelmed. Automating auxiliary tasks while adhering to best practices of human oversight becomes a necessity with enormous potential to make AI better and more trustworthy.

Modern machine learning can be unpredictable, with a focus on associations. This emphasis often produces unwanted results. Introducing causal notions lets us interpret the entire ML pipeline—from relations to results—mitigating catastrophic situations in sensitive fields.

We specialise in modelling, predicting, and explaining events, both observable and latent. Our team uncovers and explains the complex processes where different factors impact health, safety, and economic stability. By advancing data preprocessing, spatiotemporal modeling, and explainability, we enable businesses to drive data-informed decisions across risk assessment, strategic planning, and beyond.

Property owners and operators often lack essential information about the equipment installed in their buildings, which is vital for optimal operation and maintenance. Extracting this information from various scanned documents, including tables and text, typically requires significant manual effort. Our approach significantly reduces this workload.

DSA excels at extracting information from small or sparse datasets by leveraging advanced techniques in missing data imputation, transfer learning, and knowledge-intensive machine learning. Our work exploits knowledge graphs, physics-informed machine learning, and other expert knowledge integration techniques to enhance predictive accuracy and robustness, even when data is limited or incomplete.

Why rely on a black box model and hope for the best when you can enrich it with knowledge from physics or chemistry?