Is Reinforcement Learning all you need?

Speaker : Lorenzo Maggi (Nokia Bell Las)
Check our website : www.lincs.fr
Abstract:
When attacking a new problem, the algorithm designer typically follows 3 main steps:
1) rule out options that are unlikely to work well
2) test the handful of options that made the first cut
3) pick the best one and polish it until it shines
When reporting her/his work, the algorithm designer will proudly focus on step 3), briefly mention 2) and likely sweep 1) under the carpet. Yet, skimming alternatives off is a crucial step, that inevitably impacts (positively or negatively) months of hard work on testing and polishing.
I would like to lift the veil on step 1) and start sharing some subjective, but hopefully relatable, insights on how to navigate through the staggering amount of options available to the algorithm designer to solve a sequential decision-making problem, notably in the field of telecommunications.
Based on a recent blog post mlwithouttears...