Ready to learn · Level 4 · 6 points
Machine Learning and Vision
Learning from data and honest evaluation, linear and logistic models, classifiers, PCA and clustering, neural networks and backpropagation, 2-D image filtering, segmentation, convolutional networks, deep learning for signals, and squeezing a model onto a microcontroller.
Prerequisites: Embedded Systems, Multivariable Calculus, Mathematical Methods
What you'll learn (22 topics)
- Events and probability
- Point estimation: moments, maximum likelihood, bias and variance
- Learning from data and evaluation
- Least-squares regression
- Gradient and directional derivatives
- Linear and logistic models
- Detection in noise: matched filter and bit error rate
- Bayes, neighbour and tree classifiers
- Joint distributions and correlation
- Eigenvalues and eigenvectors
- Features, PCA and clustering
- The multivariable chain rule
- Neural networks and backpropagation
- Discrete convolution and impulse copies
- Image filtering and edges
- Thresholding, morphology and segmentation
- Convolutional networks
- Windowing and spectral estimation
- Deep learning for signals
- Low-power design: duty cycling, battery life, CMOS power and sleep break-even
- Real-time basics: polling, utilisation, response time and state machines
- Models on microcontrollers
