This technique focuses on getting an agent to decide

[sujonkumar6300]

Reinforcement learning.For itself which actions or paths to take to maximize the benefit obtained. For example, consider how we teach a dog a trick: we cannot give it a detailed specification of everything we want, but we can reward/punish behaviors, letting the animal deduce what it did to get the reward or punishment. We can use a similar method to train computers. Although this is a field in which there is currently great emphasis, there is still a long way to go. The type of problems that are solved are still very limited. For example, we can use it to learn to play board games such as chess or Go —in fact, reinforcement learning was used to train the AlphaGo program [6]— in which all actions are visually observable, but there is still a long way to go from that to being able to solve problems in complex environments.

Attention-based networks. Another promising area of ​​research is the incorporation of attention austria consumer email list mechanisms into neural networks. Broadly speaking, attention mechanisms consist of being able to focus only on the features of an image that have relevant information to solve the problem at hand. Attention mechanisms are not something new and have been used for some time in image recognition, but recently they have begun to be used in conjunction with recurrent neural networks (RNN), typically used for natural language processing. The results are very positive, yielding about 100 times fewer operations required to achieve the same performance [7] than traditional RNNs.