Essex joins world's elite at top AI conference

Essex researchers are joining the world's AI elite in Sydney, Australia, this December.

The Essex research team will present its work at one of the world’s most prestigious artificial intelligence conferences, alongside leading universities and technology giants including Google, Microsoft and Meta.

A paper from the Human-Centred Artificial Intelligence Lab at Essex has been accepted to the main track of Neural Information Processing Systems (NeurIPS) 2026, one of the most high-profile and competitive conferences in AI research.

The research explores why existing dynamic feature selection (DFS) methods used in AI systems do not always find the best solution and proposes a new hypernetwork-based method, Hyper-DFS, to overcome these limitations. This could help make AI systems more efficient and adaptable.

Dr Javier Andreu-Perez, from the School of Computer Science and Electronic Engineering and chair of the HCAI-Essex Lab, said: “To have our research accepted for main track paper publication at NeurIPS is a fantastic achievement for our research group.

“To be presenting alongside the world’s elite in AI research is a real testament to the work being carried out at Essex.”

The researchers have found that Hyper-DFS performs far better than existing systems when faced with combinations of information it had never seen before, with potential applications in areas such as medical diagnosis, where gathering information costs time and money.

Dr Javier Fumanal-Idocin, a YUFE MSCA Postdoc Fellow at the School of Computer Science and Electronic Engineering is first author on the research. He said: “Hypernetworks are a special kind of neural networks that generate other neural networks.

“They are great for adapting our solutions to different contexts, which is why in this paper we study what happens when we have incomplete data.

“We show that using hypernetworks can create other neural networks that adapt very well to every individual situation.”

YUFE MSCA Postdoc Fellows are an initiative led by the Young Universities for the Future of Europe, with talented researchers completing research training at YUFE partner universities including Essex.