Variational Autoencoder - By EugenioTL - Own work, CC BY-SA 4.0, link here
Right??
How to train an ML model
Bias in Machine Learning
Source: [1] (see sources slide.)
Source: [1] (see sources slide.)
Source: [1] (see sources slide.)
How to preprocess data for an ML model?
256 x 256
)Examples of different data augementation techniques. Source: [2]
What is learning?
1002 + 2003 = 3005
1 + 1 = 2
1 + 1 = 3
\[ Y = f(x_1, x_2, x_3, ...) \]
How do we create an ML model?
A diagram of a dense layer with 5 neurons, connected to another dense layer with 3 neurons.
(256, 256, 3)
to (128, 128, 3)
(256, 256)
to (65536)
(256, 256)
to a Dense layer with 256 neurons: 16.8 million parameters!(128, 128, 3)
How to train an ML model?
Number of epochs plotted against loss in an ideal situation.
Any questions?
Thank you to Dr Sam Ebbs (radiology registrar) for reviewing the slides for this seminar.
[1] - Norori N, Hu Q, Aellen FM, Faraci FD, Tzovara A. Addressing bias in big data and AI for health care: A call for open science. Patterns (N Y). 2021;2(10):100347. Published 2021 Oct 8. doi:10.1016/j.patter.2021.100347
[2] - Kermany, Daniel; Zhang, Kang; Goldbaum, Michael (2018), “Labeled Optical Coherence Tomography (OCT) and Chest X-Ray Images for Classification”, Mendeley Data, V2, doi: 10.17632/rscbjbr9sj.2. Data adapted under CC BY 4.0 Licence.