The following speakers have graciously agreed to give public lecture, keynotes, and panel discussion at ALTA 2026. Other speakers will be anounced soon!

Prof. Eduard Hovy - Executive Director of Melbourne Connect, University of Melbourne, School of Computing and Information Systems

Exploring the Representation Space of LLMs

We know that LLMs can be trained to capture simple facts, images, inference rules, generalized concepts, sentiment values, formatting, stylistic features, social/pragmatic features, and a host of other abstractions, many of which we can only vaguely articulate. How is all this represented? Is the representation space ‘linear’ (whatever precisely that means), as the Linear Representation Hypothesis postulates? Or is it approximately isotropic, or mainly linear but locally anisotropic in spots? Given that concepts are represented as complex patterns in LLMs, and given that multiple concept facets superpose within a neuron, it is impossible to even identify exactly where and how simple concepts are represented, or to easily perform effective ablation. So how can one investigate these questions? And further, how would one prove that one’s claims are valid? I describe some explorations that students, colleagues, and I have been conducting over the past 2 years.

Biography

Eduard Hovy is the Executive Director of Melbourne Connect (a research and tech transfer centre at the University of Melbourne), a professor at the University of Melbourne’s School of Computing and Information Systems, and an adjunct professor at the Language Technologies Institute in the School of Computer Science at Carnegie Mellon University. In 2020–22 he served as Program Manager in DARPA’s Information Innovation Office (I2O), where he managed programs in Natural Language Technology and Data Analytics. Dr. Hovy completed a Ph.D. in Computer Science (Artificial Intelligence) at Yale University in 1987 and was awarded honorary doctorates from the National Distance Education University (UNED) in Madrid in 2013 and the University of Antwerp in 2015, as well as several other international awards. He is one of the initial 17 Fellows of the Association for Computational Linguistics (ACL) and is also a Fellow of the Association for the Advancement of Artificial Intelligence (AAAI).

Dr. Hovy’s research focuses on computational semantics of language and addresses various areas in Natural Language Processing and Data Analytics, including in-depth machine reading of text, information extraction, automated text summarization, question answering, the semi-automated construction of large lexicons and ontologies, and machine translation. In early 2026 his Google h-index was 113, with over 75,000 citations.

Dr. Hovy is the author or co-editor of eight books and over 400 technical articles and is a popular invited speaker. He regularly co-taught Ph.D.-level courses and has served on Advisory and Review Boards for both research institutes and funding organizations in Germany, Italy, Netherlands, Ireland, Singapore, and the USA.

From 2003 to 2015 he was co-Director of Research for the Department of Homeland Security’s Center of Excellence for Command, Control, and Interoperability Data Analytics, a distributed cooperation of 17 universities. In 2001 Dr. Hovy served as President of the international Association of Computational Linguistics (ACL), in 2001–03 as President of the International Association of Machine Translation (IAMT), and in 2010–11 as President of the Digital Government Society (DGS).