Perception, Multimodal Learning, and Reasoning

Silicon Austria Labs (SAL) is Austria's top research center for Electronics and Software Based Systems (ESBS). At three locations (Graz, Villach, Linz), around 350 employees are working on future-oriented solutions for industrial production, health, energy, mobility, safety and more.

There are three research units conduct AI-related research and evaluations. Embedded AI (eAI, Linz) focuses on (1) Machine Learning and Enhanced Signal Processing, (2) Neuromorphic architectures and Chip Design, and (3) Realtime Embedded AI systems and Model Compression. Collaborative Perception and Learning (CPL, Graz) investigates (1) Model-based and Symbolic AI, (2) Federated Learning and Multi-agent Reinforcement Learning, and (3) Explainable AI and Hierarchical Models. Trustworthy & Adaptive Computing (TAC, Graz) studies (1) Model-based Diagnosis and Reasoning (w.r.t. AI safety), (2) Efficient Trustworthy Computation on the Edge, and (3) Resilience to Disturbances, Uncertainty, and Drift/Novelty. At eAI, we cover research topics in a wide range, addressing perception from a variety of sensory signals and unstructured data (e.g., radar signals; RGB, infrared, ultrasound, and thermal imaging; audio, text, etc.) including multimodal fusion. We also leverage Large Language Models (LLMs) and Vision Language Models (VLMs) with chain-of-thought and reinforcement learning from human feedback (RLHF) for complex tasks.

We are happy to join a consortium with our extensive experiences on embedded AI and efficient computation, hardware-AI, and distributed learning.