Course Topics
-
AI Foundations & Responsible AI
History, human oversight, ethics, privacy, and governance.
-
Neural Networks & Computer Vision
Perceptrons, neural networks, CNNs, and image classification.
-
Reinforcement Learning
Actions, rewards, policies, exploration, and Q-learning.
-
AI Infrastructure, Energy & Society
GPUs, data centers, energy demands, and societal impact.
-
Machine Learning, NLP & LLMs
Core concepts, language applications, capabilities, and limitations.
-
Reasoning, Assistants & Agents
Reasoning models, evaluation, tool use, MCP, and workflows.
Explore how a simple artificial neuron turns several recommendations into a Yes or No decision. Change each response and trust weight, then watch the evidence combine.
Build a network one concept at a time, watch information flow, and see how training changes influence—without equations or code.
Explore how convolutional neural networks (CNN) identify visual patterns by applying filters and combining simple features into increasingly complex representations.
See how a CNN can score well overall while repeatedly missing rare classes, then test how balanced sampling changes the result.
Learn why Q-tables work, calculate one update at a time, and solve guided worksheet-inspired challenges.
Bridge Q-tables and deep reinforcement learning by tracing Breakout frames through a CNN, replay memory, and one action-value update.
Predict a robot’s move, watch it learn a grid world through trial and error, and explain how rewards shape its policy.
Type Python to inspect fish data, build visualizations, and turn patterns into careful evidence-based claims.
Train regression models, protect test data, compare errors, and see when model complexity becomes overfitting.
Build a loan prediction model, check its mistakes, and learn why accuracy alone is not enough.
Hide species labels, cluster fish by measurement, and test why discovered groups are not automatically true categories.
Follow language from tokens and vectors into semantic space, then watch a recurrent neural network build context one word at a time.
Explore n-grams, count word patterns, predict the next word, and generate text. Later chapters preview tokens, context, and attention.