Study sheet: AI and Machine Learning Fundamentals

📋 Course Outline

1. Artificial Intelligence Foundations
2. Machine Learning Principles
3. Core Machine Learning Types
4. Supervised and Unsupervised Learning
5. Reinforcement and Deep Learning
6. Predictive and Generative AI
7. AI Applications in Networks
8. Cisco Catalyst Center AI Features

📖 1. Artificial Intelligence Foundations

🔑 Key Concepts & Definitions

• Artificial intelligence : uses computers to simulate intelligence, enabling behaviors typically associated with humans, including recognizing patterns, learning, making decisions, and solving problems

★ Must-know

📌 AI has grown in importance because of increased computing power, the availability of big data, and breakthroughs in AI research.

Further detail

• Examples of AI applications include: virtual assistants such as Siri, Alexa, and Google Assistant, recommendation systems such as Netflix, YouTube, and Amazon product recommendations, self-driving cars and robotics such as Tesla FSD and Waymo, chatbots such as ChatGPT and virtual concierges, game analysis systems such as Stockfish and AlphaGo

💡 Memory Hook

AI: recognize, learn, decide, solve

📖 2. Machine Learning Principles

🔑 Key Concepts & Definitions

• Machine learning : is a subset of AI that enables computers to learn from data and improve without explicit programming

★ Must-know

• Machine-learning algorithms identify patterns in input data and use those patterns to make predictions or decisions on new data.

Further detail

• Machine-learning applications include:
  • email spam filtering
  • personalized product recommendations
  • fraud detection in banking
  • natural language processing

💡 Memory Hook

AI is the wider field; ML learns from data instead of following only hard-coded instructions

📖 3. Core Machine Learning Types

🔑 Key Concepts & Definitions

• Supervised learning : trains a model on labeled data, where correct answers are provided, so it can make predictions or classifications on new data
• Unsupervised learning : gives a model unlabeled data and asks it to find patterns, relationships, or groupings within the data
• Reinforcement learning : trains a model through interaction with an environment, using rewards or penalties for actions to maximize performance over time
• Deep learning : is a specialized subset of machine learning that uses multilayered neural networks to process large datasets and perform complex tasks

💡 Memory Hook

Labels → patterns → rewards → layers

📖 4. Supervised and Unsupervised Learning

★ Must-know

• In supervised learning, the model examines labeled examples, learns the relationship between each input and its label, and classifies unseen data.

• Unsupervised learning groups similar data points into clusters based on shared features and can reveal hidden patterns without human supervision.

Further detail

📌 Supervised learning is highly accurate when labeled data is available, but it requires large labeled datasets that can be expensive and time-consuming to create.

📌 Unsupervised learning does not require labeled data, but its results require interpretation and labeling and are generally less accurate.

💡 Memory Hook

Supervised learning uses labels; unsupervised learning discovers structure without labels

📖 5. Reinforcement and Deep Learning

🔑 Key Concepts & Definitions

• Artificial neural network : is a computational model inspired by how biological neural networks such as the human brain process information

★ Must-know

• 🔄 Reinforcement learning proceeds by: interacting with an environment, taking an action, receiving a reward or penalty, learning which actions produce the best results

• 🔄 A deep-learning network processes data through: the input layer, multiple hidden layers that extract increasingly abstract features, the output layer

Further detail

• Reinforcement-learning applications include: self-driving cars, game AI such as Chess and Go, robotics tasks such as walking or picking up objects

📌 Deep learning handles large unstructured datasets and can achieve state-of-the-art performance in image recognition, natural language processing, and autonomous driving, but it is resource intensive and can be a difficult-to-interpret black box.

💡 Memory Hook

Reinforcement learning optimizes rewards through action; deep learning extracts features through layers

📖 6. Predictive and Generative AI

🔑 Key Concepts & Definitions

• Predictive AI : uses machine learning to analyze historical data and predict future outcomes, trends, or events
• Generative AI : uses machine learning to learn patterns from existing data and create new content such as text, images, or audio

📝 Essential Points

• Predictive AI applications include:

  • healthcare outcome prediction
  • network security anomaly detection
  • traffic congestion prediction
  • business forecasting
  • weather forecasting

• Generative AI applications include: text generation with ChatGPT, Gemini, and Copilot, image generation with Midjourney and DALL-E, video generation with Sora and Veo 2

📌 Generative AI can automate creative content creation, but it carries risks of misuse such as deepfakes and plagiarism, and its outputs can contain hallucinations or reflect poor training material.

💡 Memory Hook

Predictive AI forecasts outcomes; generative AI creates new content

📖 7. AI Applications in Networks

★ Must-know

• Predictive AI can forecast network traffic, detect security threats, and anticipate hardware failures through predictive maintenance.

• Generative AI can support networks by creating:

  • network documentation
  • device configurations
  • network designs
  • troubleshooting solutions
  • automation scripts such as Python scripts

Further detail

• Predictive AI analyzes historical and current network data to identify patterns that support bandwidth optimization, threat mitigation, and reduced downtime.

💡 Memory Hook

Historical and live network data → forecasting, threat detection, and predictive maintenance

📖 8. Cisco Catalyst Center AI Features

🔑 Key Concepts & Definitions

• AI Network Analytics : AI Network Analytics establishes baseline network behavior, provides optimization insights and recommendations, and continuously monitors the network to detect and predict anomalies.
• Machine Reasoning Engine : The Machine Reasoning Engine (MRE) uses AI to perform root-cause analysis when network issues arise, suggest resolutions, or take automated corrective actions.
• AI Endpoint Analytics : AI Endpoint Analytics identifies and classifies network devices, detects unauthorized devices or unusual behavior, and automates device profiling and segmentation.
• AI-enhanced Radio Resource Management : AI-enhanced Radio Resource Management dynamically adjusts radio settings to balance wireless load, reduce interference, and improve coverage across wireless access points.

📝 Essential Points

• Cisco Catalyst Center, formerly called DNA Center, uses AI-enabled features to identify issues before they affect users, reduce resolution time, and improve network performance and security.

💡 Memory Hook

Analytics → reasoning → endpoints → radio resources

📊 Synthesis Tables

Machine Learning Types

Type	Training data or feedback	Primary purpose
Supervised learning	Labeled data	Prediction or classification
Unsupervised learning	Unlabeled data	Pattern discovery and clustering
Reinforcement learning	Rewards or penalties from an environment	Learning actions that maximize performance
Deep learning	Large datasets processed by multilayered neural networks	Complex tasks such as image recognition and NLP