Unit 1 Flashcards
Concept Learning
Definition & Example
Definition: Searching through a predefined of potential hypothesis for the hypothesis that best fits the training examples.
Example: Humans idenfity a vehicle based on specific features, forming a concept.
Machine Learning Definition & Three Key Aspects
Definition: Technology training machines for actions like predictions and recommendations.
Three key aspects: Task, Experience, and Performance.
Machine Learning Techniques
Supervised
Learning: Uses labeled data for predictions.
Example: Training a model with images tagged as “dog.”
Unsupervised Learning: Uses unlabeled data and explores hidden structures.
Example: Clustering doccuments into categories without labeled data.
Reinforcement Learning:
* Feedback-based learning for agents.
* Maximizes positive rewards.
Example: Training a computer program to win and play games.
Semi-supervised Learning::
* Intermediate technique with labeled and unlabeled data.
* Reduces the cost of the machine learning model.
Example : Training a model with limited labeled data for cost reduction.
Applications of Machine Learning
Image Recognition: Identifying objects, face recognition.
Example: Facebook’s auto tagging suggestion using face detection.
Speech Recognition: Converting voice to text.
Example: Google “Search by voice”
option.
Traffic Prediction: Google Maps predicts traffic conditions.
Example: Goople Maps predicting traffic based on real-time data.
Product Recommendations: E-commerce and entertainment.
Ex: Amazon suggesting products based on past searches.
Self-driving Cars: Machine Learning in Autonomous Vehicles.
Ex: Tesla’s self-driving car using unsupervised learning.
Email Spam Filtering: Machine Learning for spam detection.
More Applications of Machine Learning
Stock Market Trading: Predicting market trends.
Medical Diagnosis: Detecting diseases using ML.
Virtual Personal Assistant: AI assistants like Siri and Alexa.
Online Fraud Detection: Securing online transactions with ML.
Aspects of Designing a Learning Algorithm and Example
Machine Learning Enables: Automatic learning from data, improvements with experience, and prediction without explicit programming.
Example: Driverless cars learning to drive based on based on given data.
Designing a Learning System Steps
Choose Training Experience: Impactful data for success.
Choose the Target Function: Describe the task.
Choose Representaion: Optimize move representaion.
Choose Function Approximation: Learn from examples.
Final Design: Iterative process for stystem improvement.
Inductive Learning Algorithm(ILA)
- Generate classification iteratively.
- Overcome pitfalls of previous algorithms.
- Steps: Divide, Initialize, Count, Mark, Add Rule.
