![]() ![]() ![]() The simplest forms of agent - environment - objective function combinations are easily and readily extracted from the world of games, so it's no wonder that RL research has its many tappings in this form of pastime. Success is tracked and measured by some method of scoring, so the third key element in the RL vernacular is an objective function - that is, a way to convert the perceived state of the environment into a score metric that the agent is trying to maximize. The agent may start with zero knowledge about its environment, or some simple rules can be coded into its memory.įrom there on, the job of the agent is, through trial and error, learn how the environment works and eventually figure out a way to "win". In its core, reinforcement learning refers to a subset of machine learning where an entity capable of some decision making on its actions (called an agent) is released into a world (called an environment - but it's more often very limited, and not our world per se). I am liberally referring to people much smarter than I am, so if I botched anything badly, all the errors are on me. However, before jumping headfirst into the project, let's review what we know about those basics. I've actually long wanted to learn about RL, but apart from reading some simple basics about it, the idea never made past applying it myself. We are talking about Reinforcement Learning (RL). As if to celebrate our WeChat conquest, this one is also (mainly) about games, although its applications can and should extend far broader. So with that in mind, let's embark on yet another side project. What I realized during this period is that I have to work even harder and improve even more to achieve what I'm aiming for and not let down those kind people who gave me opportunities time and again. Part of the original question in the WeChat AI discussion group. Who would have thought? It's both a humbling and sobering experience, as the road certainly hasn't been as straightforward and rosy as it might have seem just by looking at these social media posts spaced months, even a year apart. I actually got to know the folks at Clobotics in a very serendipitous way, working on my second hobby AI project on beating a casual jumping game featured on WeChat.Īnd it was on a WeChat discussion group about AI where I posted some problems that I encountered doing the project that I received a helping hand from none other than the founder and CEO of the company. ![]() Looking back these couple of years, especially my first ever article talking about the career jump from a film and animation producer into the world of AI/ML, it certainly felt like a dream that is coming true little by little. Departing from my previous job at a cross-border e-commerce company, where I started the whole Applied AI/ML article writing shenanigan, and moving back to my family in Shanghai was a welcome change, although I will miss the cozy small team of fellow Finnish colleagues back in Shenzhen. So maybe a bit of housekeeping is in order to update what's been happening in the meantime.Ī bit over half a year ago, I joined a Computer Vision (CV) and ML startup Clobotics, based in Shanghai. It has been quite a long pause (over a year, actually) since my previous Artificial Intelligence / Machine Learning (AI/ML) post. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |