Awenyx

What is Reinforcement Learning?

Reinforcement Learning is a discipline based on autonomous and adaptive decision making in complex environments.

Reinforcement Learning, as a new and disruptive field, is in rapid evolution. RL theory and algorithms are very promising, however they cannot be effectively applied to practical fields where the state-space is very large due to the known curse of dimensionality.

Starting from the main concept of reinforcement learning, we have overcome the limitations of this approach by developing a set of proprietary algorithms, named Advanced Learning, aimed at accelerating self-learning even in presence of a vast number of variables/dimensions.

We are working to expand the concept of self-learning to a plethora of new technical areas like the development of any product or service, the determination of the best neural network architecture, the path to the best personalized financial portfolio, the synthesis of new pharmaceutical molecules, the design of new antennas for communication systems, the improvement of human diagnostics and much more.

How does Reinforcement Learning work?

The Reinforcement Learning problem involves an agent exploring an unknown environment to achieve a goal. RL is based on the hypothesis that all goals can be described by the maximization of expected cumulative reward. The agent must learn to sense and perturb the state of the environment using its actions to derive maximal reward. The formal framework for RL borrows from the problem of optimal control of Markov Decision Processes (MDP).

The main elements of an RL system are:

  1. The agent or the learner
  2. The environment the agent interacts with
  3. The policy that the agent follows to take actions
  4. The reward signal that the agent observes upon taking actions

The agent employs trial and error to come up with a solution to the problem. The agent gets either rewards or penalties for the actions it performs. Its goal is to maximize the total reward.
It’s up to the agent to figure out how to perform the task to maximize the reward, starting from totally random trials and finishing with sophisticated tactics and superhuman skills.

Why is Reinforcement Learning different from common AI?

Neural Networks are the low hanging fruits of AI and they allow to achieve great results in many applications. However, these approaches have the drawback to be passive AI platforms and can only learn from existing examples. A Reinforcement Learning System instead exploits a different class of algorithms, where an agent actions create a reaction of the environment, and in this way, RL can interact actively with it. This means that RL learning is unstructured and trial & error based, which is essential to emulate human behavior in real world applications.

Here are the important characteristics of reinforcement learning:

  • There is no supervisor, only a  reward signal

  • Sequential decision making

  • Time plays a crucial role in Reinforcement problems

  • Feedback can be delayed 

  • Agent’s actions determine the subsequent data it receives