Late last year I started running into more companies using reinforcement learning (RL). Inspired by some of the things I was hearing about, early this year I wrote about emerging RL use cases in simulation & optimization, as well as examples of RL in recommendation and personalization systems. With the global pandemic taking a toll on the economy, how has demand for RL played out? Based on job postings in a few technology hubs in the U.S., demand for reinforcement learning has been quite stable. In comparison the number of postings1 containing the phrase “deep learning” (DL) declined 11% over the same period (Dec/2019 compared to Dec/2020):
The number of job postings that mention RL is about one-seventh the number of postings that mention DL. RL is arguably even more challenging to use than DL and thus more likely to be tabled during challenging economic periods. So it’s impressive that we didn’t see a comparable decline over the past year. Based on the growing number of researchers using RL, I expect that 2021 will bring more practical examples and tools that are accessible to developers.
Bonus chart: Looking at broader categories – “python”, “machine learning”, “data analytics” – this past year has seen double-digit declines in job postings. For example, the number of job postings that mention “machine learning” declined 23% (Dec/2019 compared to Dec/2020):
- One Simple Chart: which sectors are using reinforcement learning
- One Simple Graphic: companies that offer deep neural network accelerators
- Enterprise Applications of Reinforcement Learning: Recommenders and Simulation Modeling
(1) These charts pertain to raw number of job postings: companies may post a job opening multiple times; the charts in this post do not account for duplicate postings.