Autonomous systems are intelligent systems with autonomous decision-making ability. Polixir creates systems that understand the world, learn by looking into billions of different futures, and take the decision that ensures the best future.
Fit wide-range decision scenarios without human supervision
Decisions are optimal, robust, and interpretable
Free of trial-and-errors lead to low-cost deployments
Polixir technology reconstructs virtual-worlds from historical data. Even with limited data, the built virtual-world is able to generalize to unseen situations, which is essential for looking into the future.
Efficient Reinforcement Learning
Polixir continuously advances reinforcement learning algorithms. New algorithms learn with much fewer samples in complex and large-scale environments, leading to fast delivery of solution models.
Polixir incorporates adversarial attacking to ensure the robustness of solution models in the wild. The solution models are able to perform stably in extreme corner cases, even though these cases are unseen.
Polixir delivers solution models that can actively adapt to the deployment environments. Adaptive solution models keep sensing the environment and maintain the maximum performance in the changing real-world.