Scribd
Upload
18 views12 pages

Part 1

Uploaded by

d
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
18 views12 pages

Part 1

Uploaded by

d
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 12

Module 1: Fundamentals of Machine Learning and

Artificial Intelligence
• The machine learning process
• Training data: labeled, unlabeled, structured, unstructured
• ML algorithm: supervised, unsupervised, reinforcement
• Model: batch inferencing, real-time inferencing
• Deep learning
• Use cases: computer vision, natural language processing (NLP)

1
Module 1: Fundamentals of Machine Learning and
Artificial Intelligence
• Generative AI
• Foundation models
• Lifecycle: data selection, pre-training, optimization, evaluation, deployment, feedback
• Common types: LLMs, diffusion models, multimodal models, GANs, VAEs
• Model optimization
• Prompt engineering: instructions, context, input data, output indicator
• Fine-tuning: instruction fine-tuning, RLHF
• Retrieval Augmented Generation (RAG)

2
Module 1: Fundamentals of Machine Learning and
Artificial Intelligence
• AWS AI/ML services
• Amazon SageMaker
• AI/ML Services: Amazon Comprehend, Translate, Textract, Lex, Polly, Transcribe,
Rekognition, Kendra, Personalize
• Generative AI Services: Amazon SageMaker JumpStart, Amazon Bedrock,
Amazon Q, Amazon Q Developer

3
Module 2: Exploring Artificial Intelligence Use Cases
and Applications
• Artificial intelligence
• …and it’s relationship with machine learning, deep learning, and generative AI
• Real-world use cases and applications of AI
• Machine learning
• When is AI and ML an appropriate solution, and when is it not?
• Use cases of supervised, unsupervised, and reinforcement learning

4
Module 2: Exploring Artificial Intelligence Use Cases
and Applications
• Generative AI
• Capabilities: adaptability, responsiveness, simplicity, creativity and exploration,
data efficiency, personalization, and scalability
• Challenges: regulatory, social risks, data security and privacy, toxicity,
hallucinations, interpretability, nondeterminism
• Model selection considerations: model types, performance, capabilities,
constraints, compliance, cost
• Business metrics: for example, user satisfaction, average revenue per user, cross-
domain performance, conversion rate, efficiency

5
Module 3: Developing Machine Learning Solutions
• The machine learning development lifecycle / pipeline:

6
Module 3: Developing Machine Learning Solutions
• Tools to assist with stages in the ML lifecycle:
• Data preparation: Amazon SageMaker Data Wrangler
• Managing features: Amazon SageMaker Feature Store
• Model training: Amazon SageMaker, Amazon SageMaker Canvas, Amazon
SageMaker JumpStart
• Model evaluation: Amazon SageMaker Experiments, Amazon SageMaker
Automatic Model Tuning
• Deployment: Amazon SageMaker Endpoints
• Monitoring: Amazon SageMaker Model Monitor

7
Module 3: Developing Machine Learning Solutions
• Model sources
• Pre-trained models: within Amazon SageMaker JumpStart
• SageMaker building-in algorithms: for training your own models, such as linear
learner, XGBoost, MXNet, Tensorflow, etc.
• Note: you don’t need to be familiar with these algorithms
• Custom docker images: useful if SageMaker does not have the algorithms of
frameworks you wish to use built-in

8
Module 3: Developing Machine Learning Solutions
• Model evaluation concepts
• Datasets: validation, test
• Model fit: overfitting, underfitting, balanced
• Bias and variance

9
Module 3: Developing Machine Learning Solutions
• Model evaluation metrics
• Classification problems: confusion matrix, accuracy, precision, recall, F1 score,
AUC-ROC
• Regression problems: mean squared error, R squared
• Note: you are not expected to calculate these metrics, just understand how they are used

10
Module 3: Developing Machine Learning Solutions
• Model deployment
• Self-hosted API
• Managed API: Amazon SageMaker endpoints
• Real-time or batch transform
• Asynchronous
• Serverless option available

11
Module 3: Developing Machine Learning Solutions
• MLOps
• Benefits: Productivity, reliability, repeatability, auditability, and data and model
quality improvements

12

You might also like