AI Neural Network Can Perform Human-Like Reasoning
https://www.rdmag.com/article/2018/09/ai-neural-network-can-perform-human-reasoning
This article in Research and Development Magazine explains a new development in artificial intelligence technology, neural networks. Due to recent advances in research, scientists have been able to create a neural network that is able to perform “human-like reasoning procedures to answer questions about the contents of images.” Neural networks learn from data, and the software can learn and grow with more use. They are comprised of “input and output layers” as well as middle layers that “transform the input into the correct output.” This network was created by breaking the deep layers into smaller components, or modules, that each are specialized in a certain area. When the network then receives input, it is sent to the corresponding module, which sends it to the next module and so on until an answer is reached.
Although the networks described in this article are different than the traditional networks we are talking about in this class, there are many similarities. The high-level architecture of neural networks is essentially that of any network or graph we have talked about in this course so far. There are input components which have mappings to the hidden components explained in the article. Then the hidden components are all matched to the output components, and all of these components are interconnected. As more data is sent through the neural network, more connections between these components form, intertwining the network even more. I choose this article because it sheds light on the real-world applications of networks past social networks or traffic patterns. As time goes on, neural networks and artificial intelligence will become more and more prevalent in technology, and it all comes down to the combination and compilation of many simple networks.