Machine Learning --what it is Flashcards
What is Machine Learning?
Machine learning is like teaching a computer to learn by itself. Machine learning algorithms and techniques are used to make computers learn and improve from data.
1) A computer program or piece of software that learns inputs to outputs, or A to B mappings.
2) Also, a field of study that gives computers the ability to learn without being explicitly programmed. (Arthur Samuel)
What is the relation between Machine learning and AI system?
A piece of software or machine learning system that any time of day, any time of night, you can automatically input A, these properties of house, and output B.
An AI system serving dozens or hundreds of thousands of millions of users.
Think of AI as a big circle.
Inside AI, you have machine learning.
Inside machine learning, there’s a technique called neural networks, one of the most popular techniques for machine learning.
reinforcement learning, graphical models, planning, and knowledge graphs. are all tools that people use in AI
What is Data science?
The science of extracting knowledge and insights from data. data science as a field is often about using various tools to try to gain insights, to try to extract knowledge from data.
For example, Did you know that newly renovated homes have a 15% premium, and this can help you make decisions such as, given a similar square footage, do you want to build a two-bedroom or three-bedroom size in order to maximize value?’ Or, ‘Is it worth an investment to renovate a home in the hope that the renovation increases the price you can sell a house for?’ So, these would be examples of data science projects, where the output of a data science project is a set of insights that can help you make business decisions, such as what type of house to build or whether to invest in renovation.”
data science is big circle as well.
inside data science are various techniques including supervised learning, unsupervised learning,
Often the output of a data science project is often a slide deck, the PowerPoint presentation that summarizes conclusions for executives to take business actions or that summarizes conclusions for a product team to decide how to improve a website
What are artificial Neural networks?
Artificial Neural Networks (ANNs): Artificial neural networks are models inspired by the structure and function of the human brain. ANNs are computational models composed of interconnected nodes, called neurons, organized in layers. They can process and learn from data, enabling tasks like pattern recognition and decision-making.
Neural networks are a particular type of mathematical model that is capable of learning input-output mappings.
Artificial neural networks are inspired by but are not identical to the biological brain.
They are a mathematical way of implementing some functions that is often really helpful in making a computer learn input-output relationships
the mathematical model of a biological brain are actually quite unrelated, even though they happen to have the same name.
What are some examples of A to B outputs?
the input and output can be pretty much anything, depending on the application that you have in mind.
Example #1: Pattern Recognition
Imagine you have a friend named Anna who always wears a red shirt and a friend named Ben who always wears a blue shirt. One day, you notice that whenever they are together, they always play different musical instruments. You can use AB mapping to recognize the pattern: Anna + Red Shirt = Plays Guitar, Ben + Blue Shirt = Plays Drums. This pattern recognition helps AI systems identify similar patterns and make predictions based on them.
Example #2: Recommendation Systems:
Imagine you are watching a video streaming platform, and it suggests movies or shows you might enjoy based on your previous choices. The platform uses AB mapping to analyze your preferences and compare them to other users who have similar tastes. By finding patterns in the choices you and others have made, the AI system can recommend content that aligns with your interests, helping you discover new movies or shows you might like.
Example #3; Let’s say you have a housing dataset like this with the size of the house, number of bedrooms, number of bathrooms, whether the house is newly renovated . If you want to build a mobile app to help people price houses, so this would be the input A, and the price would be the output B.
What are other ways machines learn, besides A to B mappings?
Unsupervised learning
Reinforcement learning
graphical models
planning
knowledge graphs.
are tools or subfields of AI that can make computers act intelligently.
What is Unsupervised learning?
When a computer tries to find patterns or organize things by itself without someone telling it what to look for.
Example #1 : Image Clustering:
Imagine you have a collection of various animal pictures mixed together. With unsupervised learning, an AI system can analyze the images and identify similar patterns or features without any prior labels. It might group together images of cats, dogs, and birds based on their visual similarities. This clustering helps AI systems organize and categorize large amounts of data automatically.
Example #2: Anomaly Detection
Imagine you are monitoring a system for unusual behavior, such as detecting fraudulent transactions in a banking system. Unsupervised learning can be used to develop models that learn the normal patterns of transactions based on historical data. The AI system can then identify transactions that deviate significantly from these normal patterns, helping detect potential anomalies or fraud attempts.
What is reinforcement learning?
Reinforcement Learning: when a computer learns to do something better by trying and getting rewards or positive feedback when it does well. the computer improves its skills by trying different things and receiving rewards or feedback when it makes good decisions.
Cxample 1: Gaming):
In a popular video game, there is a character called “RoboBot” that can learn and improve its gameplay over time. When RoboBot defeats enemies or completes quests successfully, it receives rewards like extra points or new abilities. Through reinforcement learning, RoboBot learns which actions lead to rewards and adjusts its strategy to maximize the rewards, making it a formidable opponent in the game.
Example 2: Robotic Arm Control
In a manufacturing setting, there might be a robotic arm tasked with assembling products. Through reinforcement learning, the robotic arm can learn to perform the assembly tasks efficiently. By receiving feedback or rewards when it completes a task correctly and penalties for errors, the robotic arm adjusts its movements and actions to improve its performance over time, leading to faster and more accurate assembly processes.
What are graphical models?
Graphical Models:
pictures or diagrams that help us understand how things are connected or related to each other.
visual representations that show the relationships between different things or ideas, helping us understand how they are connected and how they influence each other.
Example#1: Weather Prediction:
In weather forecasting, AI uses graphical models to understand and predict weather patterns. By analyzing data such as temperature, humidity, wind speed, and historical weather patterns, the AI system creates graphical models that capture the relationships between different variables, enabling accurate weather predictions.
Example #2: Disease Diagnosis:
AI can learn through graphical models in medical diagnosis. For example, when determining whether a patient has a specific disease, AI systems analyze symptoms, medical history, and test results. By constructing graphical models that represent the relationships between symptoms and diseases, AI can assist doctors in making accurate diagnoses.
How does AI learn through planning?
Planning is when we think ahead and decide what we want to do and how we can do it step by step. Planning is the process of thinking about our goals and figuring out the best way to achieve them by breaking them down into smaller steps and organizing our actions.
Example #1: Route Planning
Imagine you want to go on a road trip to visit different landmarks. Before you start your journey, you use a map or a navigation app to plan the best route. The app considers factors like distance, traffic, and estimated travel time to suggest the optimal path for your trip. This planning process helps AI systems determine the most efficient way to reach a destination.
Example #2: Game Strategies:
In video games like chess or tic-tac-toe, AI can learn through planning by exploring different moves and their potential outcomes. By analyzing the game board and considering future moves and their consequences, AI systems can develop strategies to outsmart opponents and win the game.
Example #3: Vacation Planning:
Imagine you’re planning a family vacation, and you want to visit multiple attractions within a limited time frame. AI can help by learning through planning. By considering factors such as distance, opening hours, and transportation options, the AI system can suggest an optimized itinerary, ensuring you make the most of your vacation.
What are knowledge Graphs?
- Knowledge Graphs:
-special maps that show information about different things and how they are connected.Knowledge graphs are structured representations of information that show the relationships between different concepts, helping us organize and understand knowledge in a visual way.
Example #1: Voice Assistants
When you ask a voice assistant like Siri or Google Assistant a question, it uses knowledge graphs to provide you with relevant information. For example, if you ask, “Who is the first person to land on the moon?” The voice assistant can access its knowledge graph, which stores information about historical events, and provide you with the answer: “Neil Armstrong.” The knowledge graph connects different facts and concepts, allowing the AI system to understand and answer questions accurately.
I Learning Through Knowledge Graphs:
- Voice Assistants:
Voice assistants like Siri or Google Assistant use knowledge graphs to understand and respond to user queries. When you ask a question, the AI system retrieves information from its knowledge graph, which connects various facts, concepts, and relationships, enabling it to provide accurate and helpful answers. - Academic Research:
Researchers can use AI systems to explore large volumes of scientific papers and extract meaningful insights. By connecting relevant concepts, authors, and research findings through knowledge graphs, AI aids researchers in discovering new connections and advancing scientific knowledge.
Explain supervised, unsupervised and reinforcement learning
Supervised: A computer is given examples and learns from them. The computer is given a set of examples with inputs and desired outputs, and it learns to make predictions or decisions based on those examples.
A common technique where the computer learns from labeled examples.For instance, if given images of cats and dogs, the computer can learn to distinguish between them by looking at their features.
The input data and corresponding desired outputs are provided during training, allowing the model to learn the relationship between the inputs and outputs. Once trained, the model can make predictions or decisions on new, unseen data.
Unsupervised learning: where the computer finds patterns in data without any examples The computer analyzes data without any specific guidance or examples. It tries to find hidden structures or similarities in the data. This can be useful for tasks like clustering similar data points together.The algorithm analyzes the data to discover inherent similarities, groupings, or anomalies. It can be useful for tasks like clustering, dimensionality reduction, or anomaly detection.
Reinforcement learning,”where the computer learns by trying things out and getting rewards or punishments. Reinforcement learning is like learning from trial and error. The computer learns by interacting with an environment and receiving feedback in the form of rewards or punishments. It tries to maximize the rewards and improve its performance over time. By optimizing its actions to maximize the cumulative rewards, the agent improves its performance over time. Reinforcement learning has been successful in training autonomous systems and game-playing AI.
All these methods help computers become smarter and better at solving problems!
How does a machine learn from the data?
Machine learning algorithms derive knowledge and patterns from the provided data.
( IMAGE: CAT/ Not a Cat)
The algorithms adjust their internal parameters based on the data to improve their performance. This iterative process of learning from data enables AI systems to make accurate predictions, recognize patterns, or make informed decisions in various domains.
Machine learning algorithms learn from data by iteratively adjusting their internal parameters or weights based on the observed patterns in the training data. This learning process involves minimizing a loss function that measures the difference between the model’s predicted output and the true output. The model aims to generalize from the training data to make accurate predictions on unseen or future data.
What are three ways that machines learn from data?
1) Decision Trees: Decision trees are machine learning algorithms that make decisions or predictions by creating a flowchart-like structure based on features of the data. They use a series of if-else conditions to classify or regress data. Decision trees are widely used in various domains, such as finance (credit scoring) and medicine (diagnosis).
2) Support Vector Machines (SVM): SVM is a machine learning algorithm that analyzes data and builds a hyperplane to separate different classes or clusters. It is commonly used in image classification, text classification, and handwriting recognition tasks.
3) Random Forests: Random forests are ensemble learning algorithms that combine multiple decision trees to improve accuracy and robustness. Each decision tree in the random forest is trained on a subset of the data, and the final prediction is determined through a voting or averaging mechanism. Random forests are used in applications like medical diagnosis and fraud detection.
What is training data?
Training Data: Training data is a collection of input-output pairs used to train a machine learning model. It consists of labeled examples in supervised learning or unlabeled examples in unsupervised learning. The quality and diversity of training data play a crucial role in the performance and generalization of the trained model.
