Have you ever wanted to learn machine learning, but get confused by all the terms in machine learning and don't know how and where to apply it? And you are afraid what you learn is not out-dated?
Many learning resources only focus on model development, without telling you other important parts like responsible AI, model deployment and monitoring.
Personally I prefer a site that can give me the whole scope that I can know where I am in the machine learning world and what else is still missing.
If you feel the same, this is the resource gives you from the basic machine learning practice, debugging to production machine learning. Your model can really be deployed in production not just a jupyter notebook.
-
Goal:
- Understand data visualization, preprocessing
- Understand traditional machine learning, deep learning
- Understan the metrics
-
Tabular data (Regression)
- Kaggle House Price Prediction (SVR, Decision Tree)
-
Tabular data (Classification)
-
Computer Vision
-
Natural Language Processing
Think what will you need
think about pipeline
- try using tensorboard, responsible AI
- try autoML
- Data pipeline, kafka, Spark, Perfect
3. Go to production level machine learning(MLOps)
technical debt
- data engineering
- distributed training
- model/data tracking
- retraining pipeline
- kubernetes, kubeflow, tfx, seldon core
- promethues, ELK
- Sentiment Analysis
- Demand Prediction
- kafka
- Chatbot
- Android App (mediapipe)
- Video Surveillance
Machine learning algorithms like SVM, decision trees, random forest with real world challenges.
Neural network architecture and theories. Introducing loss function, activation functions, backpropagation, etc.
Data preprocessing, Exploritary Data Analysis(EDA), feature engineering, model selection and training, model inspection
Image classification, object detection, segmentation, depth estimation, etc.
NLP models and tasks.
Fairness, Reliability & Safety, Privacy and Security, Inclusiveness, Transparency, Accountability
Image captions, visual question answering