DeepSite - 🧬 RemixExploring the boundaries of AI through innovative applications and unconventional thinking.
Building machine learning models from data from the atb in Trondheim, to predict transportation needs and to optimize the buses and routes.
Combining cutting-edge ML frameworks with traditional architectures.
Demonstrating how unconventional thinking can lead to breakthroughs in applied machine learning scenarios.
Traditional machine learning approaches often follow predictable patterns. This project challenges those norms by exploring how unconventional data representations and model architectures can lead to unexpected performance improvements.
The project implements a hybrid architecture that combines traditional supervised learning with novel unsupervised techniques, creating a system that learns both from labeled data and from patterns it discovers on its own.
Brainstorming unconventional approaches to standard ML problems, identifying potential areas for creative solutions.
Building initial models with unconventional architectures, testing viability of creative approaches.
Applying non-standard preprocessing techniques and discovering hidden patterns in the data.
Iterating on the most promising approaches, optimizing performance while maintaining creative elements.
Showing faster convergence with creative approaches
Hybrid design combining multiple approaches
Unconventional features showing high predictive power