Data Science and specifically Machine Learning roles are some of the most in-demand jobs in the world because of the rapid growth of the machine learning industry with a projected valuation of $400 billion by 2023. However, there is still a huge shortage of competent individuals to satisfy the growing demands for data scientists.
This course hopes to lend a hand in bridging the gap between the demand and the supply of Machine Learning engineers.
By taking this course, you will gain the skills necessary to become a competent Machine Learning engineer. You will understand deep learning and train a simple neural network to identify objects in images as part of the practical, hands-on experience.
SATURDAYS 9AM TO 4PM
- Understand forward pass computation, the backward pass (backpropagation), Loss functions, and output functions of Deep Neural Networks.
- Explore model optimization techniques such Stochastic Gradient Descent, Adam and RMSProp.
- Understand model evaluation techniques such as MAE, MSE, Accuracy, Precision, Recall and AUC.
- Become introduced to Computer Vision techniques and architectures such as AlexNet, GoogleNet, VGG and Resnet.
- Use Transfer Learning and pre-trained architecture to train an image recognition system.
- Be comfortable with math concepts such as Linear Algebra, Geometry and Calculus.
- Be comfortable with statistical concepts such as probabilities, mean, median and variance
- Have experience with python or be willing to learn quickly.
- Introduction to Machine Learning
- Understand classification and regression algorithms
- Understand the Perceptron algorithm
- Deep Learning Project Pipeline
- The forward pass computation
- The backward pass (backpropagation)
- Loss functions
- Output functions
- Gradient Descent
- Variants of Gradient Descent
- Mean Squared and Mean Absolute Errors
- Confidence Interval
- Area under Curve (AUC)
- Confusion Matrix with Precision, Recall and F-score
- Vision operations such as Convolution, Pooling and Dropout
- Vision Project Pipeline
- Vision Architectures (AlexNet, GoogleNet, VGG, Resnet)
- Transfer Learning
- Set up GCP account with PyTorch Image
- Use transfer learning to train an image classifier
- Bonus: Deploy our model to a web page and test it