- A key aspect of achieving a working AI is efficient data preparation to meet the planned outcomes. This course explores the entire data lifecycle, focusing on strategies for collection, preparation, integration, and evaluation. Students will gain hands-on experience with data processing techniques, enabling them to understand the pivotal role data plays in AI development.
- Prerequisites: Basic understanding of Computer Vision AI or Interaction AI Course.
- Assessments: Weekly Case Studies, Final Valuation Report.
- Job Roles Applicability: AI Product Manager, AI Strategist, AI Consultant, AI Data Manager, AI Valuation Expert.
Week 1 - What is Data and Its Types
Learning OutcomeStudents should be able to understand how data is interpreted by computers, its various types, and its critical role in building intelligence.
1.1 Data types and their role in intelligence creation.1.2 Refactoring or transforming data to make it computer-appropriate.
1.3 Evaluating data readiness and correctness for different purposes
Practical Component
Demonstrate the impact of data types on AI use cases through hands-on experiments.
Week 2 - How Data is Collected and Prepared
Learning OutcomeFocus on appropriate data collection and preparation strategies.
2.1 Strategies for data collection, including generative data techniques.2.2 Labeling and preparing various data formats.
2.3 Evaluating the correctness of labeling.
Practical Component
Demonstrate how incorrect labels affect accuracy and improve accuracy with correct labels.
Week 3 - Integrating Different Forms of Data
Learning OutcomeLearn to blend various forms of data to create larger, more effective datasets.
3.1 Effects of disjoint data on AI outcomes.3.2 Techniques for blending datasets to improve algorithm intelligence.
3.3 Evaluating the effectiveness and accuracy potential of blended datasets.
Practical Componenet
Hands-on exploration of the impact of data integration on AI outcomes.
Week 4 - Assessing if Data is Right for the Goal
Learning OutcomeAssess data readiness for specific tasks and accuracy requirements.
4.1 Understanding data needs for various tasks and their labeling requirements.4.2 Task-based data preparation techniques.
4.3 Assessing if improving data can enhance accuracy.
Practical Componenet
Practical understanding of data sufficiency and its impact on algorithm accuracy.
Week 5 - When Data is Not Enough
Learning OutcomeExplore strategies for creating or adapting data when datasets are insufficient.
5.1 Generating relevant data through generative AI techniques.5.2 Using simpler algorithms to handle low-data scenarios.
5.3 Evaluating data sufficiency through indices and optimization techniques
Practical Component
Explore solutions to data scarcity, including generative data techniques and model adjustments.
Week 6 - Final Project: End-to-End Valuation of an AI System
Learning OutcomeWork on a real-world data preparation project, evaluate outcomes, and present results.
6.1 Identify, select, and label data for a specific use case, followed by accuracy assessment.6.2 Self and peer assessment of strategies and results.
6.3 Present and demonstrate the final project, including live testing and evaluation.
Practical Component
Students will present their models and demonstrate the impact of data preparation strategies.