Artificial Intelligence and Machine Learning Revision Notes Flashcards
Define Artificial Intelligence
The development of computer systems able to perform tasks normally requiring human intelligence. It follows a set of rules given by its developer, this is called an algorithm.
Define Machine Learning
Machine learning is a subset of artificial intelligence and includes the ability for a machine to learn to build and refine its own models, based on historic events.
Scope of AI
Expert Systems: Systems that make decisions in real-life situations by simulating the judgment and behavior of a human or an organization with expert-level knowledge.
Natural Language Processing (NLP): The ability of a computer program to understand human language as it is spoken and written – referred to as natural language.
Robotics: The design, construction, operation, and application of robots, often for tasks replicated or assisted by human intelligence.
Machine Vision: The capability of a computer to see, analyze, and respond to visual input, analogous to human sight.
Scope of ML
Supervised Learning; Algorithms trained on labeled data, the model is provided with the correct answers and examples during training. Use cases are classification and regression ie spam filtering or house price prediction (Algorithms are Decision Trees, SVM, Neural Networks)
Unsupervised Learning: Algorithms are left to find patterns and structures in unlabelled data. Use cases are clustering and anomaly detection ie customer segmentation and fraud detection (Algorithms are K-means and hierarchical clustering)
Key capabilities of AI
Pattern Recognition: AI can recognize complex patterns within large data sets far more efficiently than humans.
Natural Language Understanding: AI systems like chatbots or virtual assistants can interpret and respond to human language.
Automation of Routine Tasks: AI can automate repetitive tasks, improving efficiency and accuracy.
Decision Making: AI can assist or replace human decision-making in certain conditions with data-driven approaches.
Key Capabilities of ML
Predictive Analytics: ML algorithms can predict outcomes based on historical data.
Personalization: ML algorithms can tailor content, recommendations, and experiences to individual users.
Limitations of AI
Bias can affect results: a loan approval model might discriminate by gender due to bias in the data it was trained with
Errors can cause harm: autonomous vehicle experience failure and causes a collision
Who is liable for AI-driven decisions:” an innocent person is convicted of a crime based on facial recognition evidence
Limitations of ML
Data Dependency: ML models require large amounts of data for training, and the quality of this data greatly impacts performance. If the data is biased, the AI’s decisions will be too.
Transparency and Explainability: Many ML models, especially deep learning models, are often seen as “black boxes” where it’s difficult to understand how they reached a particular decision.
Dependence on Human Expertise: AI and ML still require significant human input for setting up models, providing the right data, and interpreting results.
Real world applications of AI
Healthcare:
Disease Identification and Diagnosis: AI algorithms can analyze medical images like X-rays or MRIs for early detection of diseases such as cancer.
Customer Service:
Chatbots and Virtual Assistants: AI chatbots can handle customer inquiries and provide information or assistance 24/7.
Entertainment:
Content Creation: AI can generate music, art, and even written content.
Real world applications of ML
Computational finance ie credit scoring or algorithmic training
Computational biology ie tumor detection or dna sequencing
Energy production for price and forecasting
Automotive, aerospace and manufacturing for predictive maintenance
