AI/ML Model Project Lifecycle

Contents

3. AI/ML Model Project Lifecycle#

3.1. Similar to Common Data Science Workflow#

        graph TD;
    A[Ask an Interesting Question] --> B[Get the Data];
    B --> C[Explore the Data];
    C --> D[Model the Data];
    D --> E[Communicate & Visualize Results];

    %% Indicating Iterations & Feedback Loops
    E --Refine Understanding--> A;
    C --Data Cleaning & Feature Engineering--> B;
    D --Revisit Data Exploration--> C;

3.2. Simple Lifecycle Examples#

3.3. AI/ML is Non-Linear#

3.3.1. Performance of the model and its output is not proportional to the input. Not just data input, but effort as well.#

Figure: “illustrates the uncertainty in ML development by showing the significant increase in effort but only minimal gains in model quality.”Source: Google Kaggle(AI//ML & ML competition platform) Case Study

3.4. Experimentation is key.#

try different architectures
try and tune hyperparameters
test different training techniques like data augmentation or different optimizers