PGDM department has organized a session on “Applications of Neural Network in a business context using Pytorch" on 30th Nov 2022. It was conducted by U-Liang Tang, Acting Dean and Head of Department at the School of Information and Communications Technology, HELP University Ipoh, Perak, Malaysia for PGDM Batch (2021-23) students in online mode.
He firstly introduced data science as it is a vast subject to understand containing algorithms, science, methods, systems, etc to extract knowledge and interpret the data and take a gist out of it or to analyze the trend and thus further apply the same on areas where it requires dynamic growth. He talked about the different domains such as law and governance, IT, behavioral sciences, the study of languages, arts and communication, business and economics, and many more. He talked about machine learning and other components like statistics and analytics which goes hand in hand. He also talked about the formation of neural networks in PYTORCH software, which was developed by Meta AI. The software operates on computer vision and natural language processing. He explained how machine learning algorithms can help humans to become more efficient once they learn to drive a car, check a product, picking up a product using AI and machine learning algorithms. He talked about philosophy, biology, and neural science behind the neural network. A neural network is a system of hardware and software patterned after the operation of neurons in the human brain. It builds a relationship in the form of data through a process that mimics the way the human brain operates. They are dynamic as they can adapt to the changes in output so that they provide the best possible result without changing the Input. NNs can also be put under the head of AI or artificial intelligence. It is popular in trading systems where neural networks are used to predict stock prices and maximize profit. It follows a series of algorithms that helps in finding the underlying relationship between the output and Input. A neural network consists of several neurons or nodes that are operated in parallel and arranged in layers or tiers. The first layer receives raw inputs and transfers them to the preceding layers to generate the desired output. The layers are highly connected at each node from tier N is connected to subsequent tiers N+1. There can be a series of nodes in the output layer that forms an image in a readable format. Each node carries a weight that entirely depends on values that contribute to getting the correct answers. In other words, nodes that contribute to getting the desired results are assigned higher weights in comparison to nodes that don’t. Initially, nodes are flooded with a huge chunk of data and output is told to the network in advance.