Title: Competition-Level Code Generation with AlphaCode,

Abstract:
Programming is a ubiquitous problem-solving tool. Developing AI systems that can independently generate programs or assist programmers can change the paradigm of programming. Recent large-scale language models have demonstrated an impressive ability to generate code, and they are now able to complete simple programming tasks.

However, these models are rather weak when evaluated on more complex, unseen problems that require problem-solving skills beyond simply translating instructions into code. For example, competitive programming problems which require an understanding of algorithms and complex natural language remain extremely challenging. To address this gap, we introduce AlphaCode, a system for code generation that can create novel solutions to these problems that require deeper reasoning. In simulated evaluations on recent programming competitions on the Codeforces platform, AlphaCode achieved on average a ranking of the top 54.3% in competitions with more than 5,000 participants. Today, I will present the key ideas of the AlphaCode system, and insights for tackling the competitive programming problem.

Bio:
Junyoung Chung is a staff research scientist at DeepMind. He received his PhD degree from the University of Montreal / MILA under the supervision of Professor Yoshua Bengio. He contributed to large-scale projects at DeepMind such as AlphaStar and AlphaCode. His research interests include various topics in deep learning, natural language processing and program synthesis.