Invited Talks

Introduces a geometric framework for multimodal alignment that replaces pairwise similarity with higher-order geometric consistency for multimodal generative AI.

This talk presents a multimodal alignment framework that jointly models temporal, spatial, and semantic consistency to enable controllable audio-visual generation for creative content creation.

This talk presents hypercomplex machine learning as a geometric framework for modeling 3D audio signals, enabling advanced spatial audio processing and immersive audio applications.

This invited talk as a panelist discusses how large language models can foster strategic thinking, creativity, and personalized learning, reshaping signal processing education beyond answer-driven instruction.

This invited lecture explores how generative AI is transforming healthcare and life sciences, from precision diagnostics and personalized medicine to drug discovery and biological research.

This invited talk within a panel discusses the growing synergy between artificial intelligence and electrical engineering, highlighting how engineering principles can inspire next-generation AI while AI reshapes the future of electrical systems.

This talk presents a generative AI framework for semantic audio communication, enabling robust audio transmission by reconstructing signals from their semantic representations rather than the transmitted bitstream.

This talk presents a unifying perspective on hypercomplex deep learning, explaining how inductive biases enable more effective learning from multidimensional data.

This invited tutorial introduces the foundations of generative AI and explores how modern generative models are transforming information and communication technologies, from multimedia synthesis to immersive digital environments.

This talk presents an information-theoretic perspective on quaternion deep generative models, showing how hypercomplex representations improve latent learning for multidimensional signals.

This talk deals with the importance of data-centric AI for audio applications, focusing on dataset design, data collection, and benchmark development for modern AI audio systems.

This talk presents the L3DAS project, a community-driven initiative advancing 3D audio AI through large-scale datasets, open benchmarks, and international machine learning challenges for immersive audio.

This talk deals with adaptive filtering techniques for noise and vibration reduction in reverberant acoustic environments.