What is the difference between Deep Learning and Reinforcement Learning?
We are often asked at Rebellion what are the differences between these two styles of Machine Learning?
Deep learning and reinforcement learning are both sub-fields of machine learning systems that learn autonomously. Deep learning uses data to train a model to make predictions from new data. Here, the goal is usually to train a computer to do as well or better than a human some task such as image recognition, speech recognition and translation, and other skills that a human is considered to be good at, without human involvement. However, unlike recognizing an image or speech problems in the financial markets may be different from typical deep learning applications, for example, select a stock that is likely to perform well in some future period.
Reinforcement learning (RL), on the other hand, utilizes a software agent to make observations and takes actions within an environment, and in return it receives rewards and its objective is to learn to act in a way that will maximize its expected long-term rewards. It offers the advantage of solving the complex sequential decision-making problems based on learning from the previous knowledge since, it provides a dynamic learning against the changing environment and no need to pre-determined model of environment.
As a result, the difference is that deep learning is learning from a training set and then applying that learning to a new data set, while reinforcement learning is dynamically learning by adjusting actions based in continuous feedback to maximize a reward, which makes it more suitable for financial applications.
However, deep learning and reinforcement learning aren’t mutually exclusive. In fact, one might use deep learning in a reinforcement learning system, which is referred to as deep reinforcement learning.
Read more from RebellionResearch.com:
Written by Arash Doosti & Edited by Alexander Fleiss