Universal AI Program by MIT Open Learning

Python Coding - Part 02

Topics Covered:

• Work with structured datasets (CSV, Pandas) and use visualizations to derive insights for business decisions.
• Develop intuition for basic machine learning models (e.g., decision trees) and how overfitting impacts reliability.

Foundations of Data Analytics and Machine Learning

Topics Covered:

• Understand the end-to-end workflow of data analytics—from collecting data to communicating outcomes.
• Build comfort with foundational statistics and how they support evidencebased decisioning.

Supervised and Unsupervised Learning

Topics Covered:

• Learn how common prediction models (regression/classification) and clustering approaches are used in business use cases.
• Interpret model quality using simple performance measures (accuracy, AUC, etc.) and explain results confidently.

Foundations of Neural Networks

Topics Covered:

• Understand what neural networks are, why they work well for unstructured data (text/images), and where they can fail.
• Build intuition for embeddings—how AI “represents meaning” in language and visual data.

Hands-On Deep Learning (End-to-End)

Topics Covered:

• Understand how deep learning models are trained (forward/backward propagation, gradient descent) and what influences outcomes.
• Experience the end-to-end steps of training and evaluating a model using modern frameworks (PyTorch/TensorFlow) in a guided way.

Deep Learning and Computer Vision

Topics Covered:

• Understand how AI powers computer vision use cases such as detection, classification, and automation of visual inspection.
• Learn how modern approaches like CNNs and transfer learning speed up delivery and improve adoption feasibility.

Data-Driven Prescriptive AI

Topics Covered:

• Differentiate prediction (“what will happen”) vs prescription (“what should we do”) and why this matters in operations and planning.
• Learn how organizations combine ML predictions with decision logic to recommend actions at scale.

Model-Driven Prescriptive AI - Part 1 (Optimization Foundations)

Topics Covered:

• Understand how optimization models work (objectives, constraints, feasibility) and how they support enterprise decisions.
• Apply optimization intuition to real problems like routing, staffing, allocation, and resource planning.

Model-Driven Prescriptive AI - Part 2 (Advanced Optimization)

Topics Covered:

• Build comfort with advanced decision trade-offs like multiobjective optimization (cost vs speed vs quality).
• Understand why real-world optimization can be complex (nonlinearities, convex vs nonconvex) and how organizations handle it pragmatically.

Large Language Models

Topics Covered:

• Understand how LLMs work at a practical level (tokens, transformers, attention) and what drives output quality.
• Learn prompting strategies (zero-shot, few-shot) and key risks (hallucination, bias, cost) for enterprise deployment.

Generative AI, Future of Work, and Human Creativity

Topics Covered:

• Understand where GenAI creates value in knowledge work—content, ideation, design—and where human judgment remains critical.
• Build conceptual awareness of how generative systems work (diffusion models, embeddings) and how to evaluate business fit.

Multimodal AI

Topics Covered:

• Understand how modern AI combines text, images, and other inputs to improve prediction, reasoning, and automation.
• Learn key challenges such as alignment, fusion, interpretability, and feasibility when scaling multimodal enterprise use cases.

Explanation, Reasoning, and AI Ethics

Topics Covered:

• Learn how to explain AI outcomes to stakeholders using practical interpretability approaches (global/local/counterfactual).
• Understand responsible AI fundamentals—bias, fairness trade-offs, and mitigation strategies aligned with enterprise governance.

AI Explainability & Fairness

Topics Covered:

• AI explainability refers to the ability to understand, interpret, and clearly communicate how an AI system arrives at its decisions or predictions.

AI Ethics

Topics Covered:

• AI Ethics refers to the principles and guidelines that ensure artificial intelligence systems are designed, developed, and deployed in a way that is responsible, fair, transparent, and aligned with human values.