<aside> ❗ Note:
See portfolio for Product Analysis Reinforcement learning (RL) and generative adversarial networks (GANs), based on preliminary knowledge, are probably not going to be as accurate as convolutional neural networks. For that reason, only brief descriptions of RLs and GANs were given as the already-existing products were analyzed.
</aside>
What did we learn?
Why is it useful?
How is it useful?
Each algorithm has a specific ability. It has distinct features that optimize it for some tasks. CNNs are good for image analysis, RL is good for agents learning from scratch, and GAN is useful for generating data.
We are now aware that CNN is best for our application: classification of images, which is CNNs strength compared to RL and GANS.
We can create a prototype using CNN as a developed idea in the development stage of the process.