Emergent Communication Group

My main research focuses on simulating the evolution of linguistic structure using neural agents. Specifically, I want to develop methods to simulate the cultural evolution of linguistic structure through neural network agent-based computational modeling, and to explore the nature of learners' bias (whether it's statistically learned or internal cognitive constraints).

It is a familiar phenomenon: you ask the assistant on your phone to call your mother, but it calls a friend instead. I investigate how humans and Artificial Intelligence (AI) can better communicate with each other, so that these kinds of situations will no longer occur in the future.

My research explores the emergence of language universals using neural agent simulations. Currently, I’m focusing on the word order/case-marking trade-off in natural languages, which is one of the most well-known language universals. I have developed a cutting-edge framework called NeLLCom, which stands for Neural-agent Language Learning and Communication. Basic model training procedures such as supervised learning and rewards fine-tuning based on communicative success are implemented there. Taking inspiration from Language Evolution research, NeLLCom also carries out various other learning paradigms, e.g. iterated learning and group communication.

I am a PhD candidate studying the roles of cooperation, social cognition and community structure in the emergence of linguistic systems, in order to make inferences about the evolution of language in early humans. To do this, I use a variety of methods, including lab experiments and computational models, drawing on findings from cognitive science, linguistics and evolutionary anthropology.