DEEP LEARNING APPLICATION IN IMAGE PROCESSING (Kỹ thuật học sâu trong xử lý ảnh) Hoàng Văn Dũng e-mail:
[email protected] Faculty of Information Technology
ISLab- Intelligent Systems Laboratory Contents • Tổng quan convnets • Huấn luyện một model dùng bộ dữ liệu nhỏ • Sử dụng các pretrained convnet • Visualizing what convnets learn • Áp dụng
ISLab- Intelligent Systems Laboratory Deep Neural Network 1. Deep neural network? 2. Why is it better than other methods on image and other types of data? Briefly answers 1. Deep neural network=> using a neural network with many layers of nodes between input and output 2. The series of layers between input & output do both feature identification and processing, ➢ Just as our brains seem to.
ISLab- Intelligent Systems Laboratory Deep Neural Network • Multi-layer neural networks have been around for 30 years. What is actually new? ➢ Good algorithms at learning the weights for networks with 1-> several hidden layers ➢ But not good at learning the weights for networks with more hidden layers ➢ Algorithms for training many-later networks
English