12/3/2021: Convolutional Neural Networks, Julio Godoy, University of Concepción

Speaker/presenter: Julio Godoy

1. i) Presentation description:

Convolutional neural networks (CNN) are a type of feedforward neural networks (NN) that have been (and continue to be) widely used for image processing tasks such as object detection and image segmentation. These types of NN are composed of three types of layers: convolutional layers, (max) pooling layers and typically fully connected layers. Convolutional layers, through the operation called convolution, extract visual patterns from an input image in various levels of abstractions using learned filters. Max pooling layers reduce the dimensionality of the processed image for more efficient processing, and fully connected layers introduce non-linearities that helps in learning to differentiate between different data classes. Beyond image processing in a supervised learning fashion, CNN have been used in fields such as signal processing, robotics and reinforcement learning.

References:

[1] Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep learning. MIT press.

[2]Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems, 25, 1097-1105.