Special designs and competition protocols

Wei-Wei Tu (1) , Adrien Pavão (2, 3)

1 4Paradigm
2 LRI - Laboratoire de Recherche en Informatique
3 TAU - TAckling the Underspecified

With the development of AI technology, many novel machine learning frameworks have been raised and applied in AI academic and industry research and business application. Organizing competitions in these areas can greatly help the research and development of related algorithms and technology. In this chapter, we explore the design of competitions in various kinds of machine learning field: supervised learning, automated machine learning, metalearning, time series analysis, reinforcement learning, adversarial learning, and using confidential data. For each of these specific competition protocol, we discuss the framework and design of the competition process. We believe this chapter can make great help to both the organizers and the participants, therefore accelerate the development of AI industry and research.

Mots clés: competition; design; supervised learning; automated machine learning; metalearning; time series analysis; reinforcement learning; adversarial learning; confidential data

Domaines: Machine Learning [stat.ML]; Informatique [cs]

Wei-Wei Tu, Adrien Pavão. Special designs and competition protocols. 2024. ⟨hal-04416565⟩