First HI Lights seminar at the University of Oulu dived into the ways human emotions and behavior can be understood from the data and algorithm aspect
The first HI Lights seminar featured an excellent international lineup of keynote speakers from diverse fields who shed the light on the path to achieving hybrid intelligence through practical tools and research. The expert key notes came from very multidisciplinary backgrounds, including human-computer interaction, machine vision, the co-evolutionary processes of humans and machines, psychiatry and behavioral sciences, and computer vision and AI.
Welcoming words were given by Guoying Zhao, vice leader of the programme. In Opening speech, Taina Pihlajaniemi, Vice Rector explained how the University of Oulu´s research profile is based on internationally recognized excellent research areas, Hybrid Intelligence being one of them as a Profilation 7 research programme.
Keynote: How to 3D model complex interactions – and also improve the fairness of AI?
The first keynote “Cross-modal understanding and generation of multimodal content”, by Professor Nicu Sebe, (Trento University), took us to animation modeling and introduced how to 3D model complex interactions. For example, with learnable game engines one can learn to play tennis like Federer. Examples of deep fake videos with singing bills made audience laugh, but also highlighted the question of fairness in AI. Could we use foundation models to propose and detect biases that are present in Text-to-Image Models?
ThemeTalks: Cognitive AI is fundamentally different from traditional AI since congnitive AI should be able to adapt to the context and environment
In session of Theme Talks from HI: Data and Algorithm PhD Qianru Xu, (University of Oulu), presented the study done in collaboration of University of Oulu and Stanford University School of Medicine. In “From Emotion AI to Cognitive AI” keynote she explained how cognitive AI is fundamentally different from traditional AI. Traditional AI is mainly based on the fed data and processed accordingly under human pre-programming, while cognitive AI should be able to adapt to the context and environment to evolve and grow over time.
ThemeTalks: Research explores whether AI can be trained to be experts in body gesture-based emotion, while AI already helps in media content and context recognition
Continuing with the same Theme Talks, PhD. Haoyu Chen, CMVS, Assistant Professor in HI Programme, University of Oulu, gave a keynote “Understand and synthesize human emotions based human behavior analysis” that shed light on the research that tries to explore whether AI can be trained to perform as experts in body gesture-based emotion, i.e., to catch micro gestures for understanding people’s hidden emotional status. After Chen, PhD. Mika Rautiainen, ( Valossa Labs), talked about how AI can help in media content recognition and understanding the context. His research team’s innovations have been integrated into Valossa’s content analysis engine, as well as the company’s emotion and expression recognition engine.
Keynote: As AI enters the sphere of the social system, there will be unexpected benefits but also harms and thus a need for human-centered Hybrid Intelligence research
In the keynote of Prof. Kaisa Väänänen, Tampere University, “Human-Centered Hybrid Intelligence: How to achieve societal impact?”, we got a wider view of AI from the human viewpoint. Al may enter the sphere of the social system, and then there will be more creative and proactive applications that lead to unexpected benefits but also harms. Prof. Sanna Järvelä, leader of the HI programme, introduced Professor Väänänen.
In the panel discussion we had our guests of honor participating, our international Scientific Advisory Board (SAB): Hirokazu Kato (Japan), Carolyn Rose (USA), Nikol Rummel (Germany), Kaisa Väänänen (Finland), and Nicu Sabe (Italy) defining Hybrid Intelligence and its role from different perspectives, interviewed by Haoyu Chen. The multidisciplinary audience was very active taking part in group discussions around Hybrid Intelligence topic after the thought-provoking panel discussion.