Machine Learning for Practitioners

  • Typ: Block-Vorlesung (BV)
  • Lehrstuhl: KIT-Fakultäten - KIT-Fakultät für Elektrotechnik und Informationstechnik
  • Semester: SS 24
  • Ort:

    ITIV Raum 339

  • Zeit:

    10.09.2024
    13:30 – 17:30 Uhr

    11.-12.09.2024
    09:00 – 13:00 Uhr

  • Dozent:

    Prof. Dr. Amlan Chakrabarti

  • Hinweis:

    Präsenz

Vortragssprache Englisch
Organisatorisches

10. September von 13:30 Uhr - 17:30 Uhr

11-12. September, jeweils von 9:00 Uhr - 13:00 Uhr

Machine Learning for Practitioners

Ziele 

The " Machine Learning for Practitioners" course is designed to elevate the skill set of participants by bridging foundational knowledge with practical, real-world applications. Over an intensive 12-hour period, attendees will revisit key machine learning concepts, delve into advanced data preprocessing and feature engineering techniques, and gain expertise in model selection, evaluation, and tuning. The course also covers sophisticated supervised and unsupervised learning methods, introduces deep learning fundamentals, and guides participants through a hands-on project to solidify their understanding. Additionally, learners will explore the essential aspects of model deployment, monitoring, and maintenance, ensuring they are well-equipped to implement and manage machine learning solutions in production environments. This course is ideal for practitioners looking to enhance their machine learning capabilities and apply them effectively to solve complex problems in various domains.

Kursinhalte

Review of Fundamental Concepts

  1. Key ML Concepts:
    • Supervised vs. Unsupervised Learning
    • Common algorithms: Linear Regression, Decision Trees, K-Means, etc.
  2. Python Libraries:
    • Scikit-Learn
    • Pandas
    • Numpy

Data Preprocessing and Feature Engineering

  1. Data Cleaning
    • Handling missing values
    • Dealing with outliers
  2. Feature Engineering:
    • Encoding categorical variables
    • Feature scaling (Normalization/Standardization)
    • Feature selection techniques

Model Selection and Evaluation

  1. Model Selection:
    • Choosing the right model for the problem
    • Understanding Bias-Variance Tradeoff
  2. Evaluation Metrics:
    • Accuracy, Precision, Recall, F1 Score
    • ROC Curve and AUC
    • Cross-validation techniques

Advanced Supervised Learning Techniques

  1. Ensemble Methods:
    • Bagging and Boosting
    • Random Forest, Gradient Boosting, XGBoost
  2. Hyperparameter Tuning:
    • Grid Search
    • Random Search
    • Bayesian Optimization

Unsupervised Learning Techniques

  1. Clustering:
    • K-Means
    • Hierarchical Clustering
    • DBSCAN
  2.    Dimensionality Reduction:
    • PCA
    • t-SNE

Deep Learning Basics

  1. Neural Networks:
    • Structure and Function
    • Activation Functions
    • Backpropagation
  2.    Frameworks:
    • Introduction to TensorFlow and Keras

Practical Application and Project

  1.   Hands-On Project:
    • Choose a dataset (e.g., from Kaggle)
    • Define the problem and approach
    • Data preprocessing, model selection, and evaluation
  2.    Code Walkthrough:
    • Live coding session for implementing a model from scratch

Deployment and Model Management

  1.    Model Deployment:
    • Introduction to Flask/Django for creating APIs
    • Using Docker for containerization
  2.    Model Monitoring and Maintenance:
    • Monitoring model performance over time
    • Retraining and updating models

Voraussetzungen

  • Basic understanding of Python programming
  • Familiarity with fundamental machine learning concepts