Introduction to Convolutional Neural Networks
Machine learning is beginning to have a major impact on software development for all platforms and industries. However a life of developing software in a traditional explicit manner is not the best training for developing machine learning algorithms. Software developers faced with problems like "What is in this image", "Is image a face and is it smiling?", or "Is this a hot dog?" would face much difficulty without a machine learning algorithm approach.
This session explains what a neural network is in simple terms. That definition is expanded to include Convolutional Neural Networks and why they are great for image recognition. A sample neural network implemented in Keras and trained with the MNIST dataset will be demostrated to recognize handwritten digits. This session will conclude with an sample iOS application using CoreML to run the neural network directly on a phone and recognize digits in real time.
You will leave this session with some ideas on how to begin to approach problems best solved with machine learning algorithms.