The 6th Annual CBAIA Conference Held in London: Spotlight on Data Accuracy in AI for Finance and Healthcare

Share This

Author: Chinese Business Gazette

London, November 9 — Experts and scholars from the UK’s academic and financial sectors gathered at the Gatsby Computational Neuroscience Unit of University College London (UCL) for the 2024 Annual Conference of the China-Britain Artificial Intelligence Association (CBAIA). The event focused on enhancing the accuracy of artificial intelligence and its successful applications across various domains.

At the thematic forum titled “Crowning AI with Data”, keynote speeches were delivered by several prominent figures, including Professor Arthur Gretton, a machine learning expert from UCL; Yangchun Chu, Chief Operating Officer of big data platform DolphinDB; Professor Peter Latham from UCL’s Gatsby Computational Neuroscience Unit; and Chris Ling, founder of fintech company AlpFin. The speakers shared insights on topics such as parameter selection and application in AI data modeling.

Professor Arthur Gretton pointed out that while AI is capable of making predictions on diverse issues — such as drug efficacy or policy impacts — it often struggles to assess the uncertainty of its own predictions accurately. In the field of healthcare, for example, AI treatment recommendations are often based on the severity of a patient’s condition: mild cases may receive drug therapy, while severe cases require surgery. Typically, drug treatment has an 85% cure rate, while surgery has a 72% rate.

These figures may lead hospital administrators to conclude that drug therapy is the superior treatment due to its higher effectiveness and lower cost. However, this inference overlooks the underlying relationship between treatment type and illness severity. Mild cases are suitable for drug therapy, whereas severe cases necessitate surgery. This example provides a crucial lesson for AI researchers and practitioners: failing to properly account for relationships between variables — such as treatment type and disease severity — can result in misleading medical data and incorrect interventions by decision-makers, leading to poor outcomes and ineffective treatment. Gretton emphasized that thoroughly examining these inter-variable relationships is essential to avoid oversimplifying complex scenarios. He cited this as one of the greatest challenges facing AI research and a key focus of his team’s ongoing work.

In his speech, DolphinDB COO Yangchun Chu shared examples of AI applications in finance. In the context of data processing and financial modeling, a typical quantitative hedge fund or trading desk must handle massive datasets — including trade instructions, company financials, and other relevant data — and use predictive models to generate trading signals. When these models perform well and accurately forecast returns, they can yield substantial profits.

The choice of the Gatsby Computational Neuroscience Unit at UCL as the venue was another highlight of the event. Founded in 1998 and supported by the Gatsby Charitable Foundation, the center was co-founded by Professor Geoffrey Hinton, a Turing Award laureate and one of the 2024 Nobel Prize winners in Physics — widely recognized as a “Godfather of AI.” The unit focuses on understanding the neural mechanisms of brain computation by integrating neuroscience, mathematics, and machine learning to explore how neural circuits process information, learn, make decisions, and generate perception and behavior. Hinton’s Nobel recognition was awarded “for foundational discoveries and inventions that advanced the use of artificial neural networks for machine learning.”

Both Professor Gretton and Professor Peter Latham are part of the elite research team at UCL’s Gatsby Unit, working alongside Professor Hinton. Gretton noted that while their current work integrating neuroscience and machine learning is still maturing, understanding the human brain remains one of the most challenging frontiers. As machine learning researchers, they are committed to pursuing this ambitious goal.

Founded in 2017, the China-Britain Artificial Intelligence Association (CBAIA) is a UK-registered non-profit organization and has grown into one of the largest Chinese tech associations in the UK. Its mission is to promote exchange among Chinese professionals in AI and foster academic and industry collaboration between China and the UK in the field of artificial intelligence.