Hardcore Data Science Explained
As the amount of data in the world grows dramatically, humanity does not have enough resources to review and label it. Algorithms and other tools for working with large amount of unlabeled data will be game-changers for businesses in the near future. One of the most powerful algorithms known today for analyzing unlabeled data is deep learning neural networks. Inspired by brain architecture and the idea of a single learning algorithm for any recognition problem, deep learning brings state of the art performance for wide variety of analytical problems such as video analysis, image analysis, audio analysis, text analysis and others.
In order to design for deep learning neural networks, you need to understand how they work and a fair amount of processing power for training. One way to get this power is with parallel processing on GPU. Training for deep learning neural networks on GPU is the most efficient and cheapest way of doing state of the art Big Data analytics today.
This class provides insight behind deep learning networks and breaks down the process of design, training and the fine tuning of these models. It also provides guidelines and a technology overview for simplifying training processes on GPU.
Note: You will need an understanding of Machine Learning.
Level : Advanced