Most AI systems learn in isolation and adapt only through retraining. But intelligence in nature works nothing like that. In this post, I share the foundational question that led me to leave a stable career and build something different.
All tagged continuous learning
Self-Learning Traffic Prediction in Los Angeles: A Continuous Learning Experiment
Years before I founded Wakeline, I let a self-learning AI watch Los Angeles traffic and adapt in real time, with no retraining. A look back at an early continuous learning experiment.
I built a self-learning AI that watched Wolfsburg traffic, adapted in real time without retraining, and ran cheaply enough to offer for free. A look back at how it learned.
Self-learning AI Movement Prediction: Beyond Airstriker Genesis to multi-directional predictions
Self-learning AI Experiment Update: Witness the system's self-learning process and evolving predictions in response to new challenges in a tower defense-style game.
Movement Prediction in Airstriker Genesis, Learned in Real Time
I’ve explored AI movement prediction in Airstriker Genesis via OpenAI Retro Gym and developed a prototype for real-time adaptation without prior training. Video showcases predictive output alongside live gameplay. Focus on refining horizontal movement detection and implementing knowledge sharing for consistent predictions.