So… How exactly is AI being used to detect COVID-19?

towards-data-science

This post was originally published by at Towards Data Science

Comparison between AI, ML & DL. Image taken from https://commons.wikimedia.org/wiki/File:AI-ML-DL.svg

Image by Author

Image by Author

Input and weights giving rise to output ŷ. Image taken from https://cs.gmu.edu/~kosecka/cs747/03_linear.pdf

Activation Function

Image by Author

Sigmoid activation function. Image taken from https://commons.wikimedia.org/wiki/File:Sigmoid-function-2.svg

Activation functions introduce non-linearities into the network. © Alexander Amini and Ava Soleimany
MIT 6.S191: Introduction to Deep Learning

Commonly used activation functions. Image taken from Sze, Vivienne, et al. “Efficient processing of deep neural networks: A tutorial and survey.” Proceedings of the IEEE 105.12 (2017): 2295–2329.

Minimizing Loss

Image by Author

Image by Author

Gradient Descent

Gradient Descent Algorithm. © Alexander Amini and Ava Soleimany
MIT 6.S191: Introduction to Deep Learning

Gradient Descent Illustrated. Image taken from https://commons.wikimedia.org/wiki/File:Gradient_descent.gif

Backpropagation

Two Layer Neural Network. © Alexander Amini and Ava Soleimany
MIT 6.S191: Introduction to Deep Learning

Image by Author

Image by Author

Impact of learning rate on training. Image taken from Krittanawong, Chayakrit, et al. “Deep learning for cardiovascular medicine: a practical primer.” European heart journal 40.25 (2019): 2058–2073.

Overview of a convolutional neural network (Image by Author)

Convolution Layer

2×2 kernel (pink) sliding over 4×4 input image (light green) to produce a 3×3 output image (dark green). Image taken from https://commons.wikimedia.org/wiki/File:Valid-padding-convolution.gif

Convolution operation with input image of size 4×3 and kernel of size 2×2 to produce a feature map of size 3×2. Image taken from Goodfellow, I., Bengio, Y., & Courville, A. (2016). Deep learning. MIT press.

Convolution operation with input image of size 6×6 and kernel of size 3×3 to produce a feature map of size 6×6. Image taken from https://commons.wikimedia.org/wiki/File:2D_Convolution_Animation.gif

Pooling Layer

Image by Author

Image by Author

Fully Connected Layer

Architecture of a convolutional neural network (Image by Author)

References

Spread the word

This post was originally published by at Towards Data Science

Related posts