Download as PDF, PPTX
Uploaded byDmitry Zinoviev
Machine Learning Basics for Dummies (no math!)
This document discusses machine learning and artificial intelligence. It defines machine learning as a subset of AI that focuses on learning, allowing computers to perform tasks without explicit instructions by relying on patterns and inferences derived from data. It also describes the machine learning process, including gathering and cleaning data, labeling the data, selecting relevant features, choosing a model type, training the model on a portion of the data, and testing the model on separate data to validate it. Rule-based medical diagnosis systems are provided as an example of early AI applications.
More Related Content
Similar to Machine Learning Basics for Dummies (no math!)
Machine Learning Basics for Dummies (no math!)
- 1.
- 2. CAS-HC225 Spring 2024 2/17 AI& Machine Learning ● Artificial Intelligence represents a broad discipline aimed at creating machines or computer programs capable of performing tasks that, if done by humans, would require some level of intelligence. These tasks include: ✔ problem-solving, ✔ understanding natural language, ✔ recognizing patterns, and ✔ learning from experience. ● Machine Learning is a subset of AI focused specifically on the aspect of learning: Learn and Adapt! ✔ Develop algorithms and statistical models that enable computers to perform tasks without using explicit instructions. ✔ Instead, they rely on patterns and inference derived from data: Give machines access to data and let them learn for themselves.
- 3. CAS-HC225 Spring 2024 3/17 Rule-BasedAI: Medical Diagnostics ● Knowledge base: Contains detailed information about various medical conditions and their associated symptoms, diagnostic procedures, and possible treatments. Developed by medical experts. ● Rule engine: Applies logical rules to the information input. ✔ IF the patient has a fever AND a sore throat THEN consider the possibility of strep throat. ● User Interface: Input patient data (symptoms, medical history, and test results). ● Diagnosis and Recommendations: Based on the rules, generate a ranked list of possible diagnoses along with recommendations for further testing or treatment options. ● MYCIN (Stanford, early 1970s), DXplain (MGH, 1984–1986)
- 4. CAS-HC225 Spring 2024 4/17 Advantages ● Consistencyand Speed: Provides consistent assessments and can process information much faster than a human. ● Accessibility: Makes expert-level knowledge more accessible, especially in underserved areas. ● Support Decision Making: Acts as a support tool for practitioners, enhancing their ability to make informed decisions.
- 5. CAS-HC225 Spring 2024 5/17 Limitations ● Lackof Flexibility: Rule-based systems can only operate within the confines of their predefined rules and cannot learn from new data or cases outside of their initial programming. ● Maintenance: The knowledge base and rules need regular updates to incorporate the latest medical research and findings, which can be resource- intensive. ● Accountability: Ethical dilemmas.
- 6. CAS-HC225 Spring 2024 6/17 MachineLearning Process ● Understanding Data ✔ Gathering relevant data. ✔ Cleaning data: Removing inaccuracies or duplicates. ✔ Labeling data: Classifying into categories. ✔ Feature selection: Choosing relevant data points. ● Choosing a Model: ✔ Regressions, decision trees, neural networks, support vector machines... ● Learning ✔ Training: The selected model is “trained” using a portion of the data set. During training, the model adjusts to better fit the data. ✔ Testing: A separate portion of the data, not seen by the model during training, is used to validate the model's accuracy and generalizability. ● Making Predictions
- 7.
- 8.
- 9.
- 10. CAS-HC225 Spring 2024 10/17 FeatureExtraction & Selection
- 11.
- 12.
- 13. CAS-HC225 Spring 2024 13/17 Supervisedvs. Unsupervised ● Supervised learning: ✔ Supply the labels to the model ✔ Movie recommendation system asks a customer whether they like a 👍 particular movie or not (or asks for a rating) 👎 ● Unsupervised learning: ✔ Do not supply the labels, let the model find the patterns itself ✔ Movie recommendation system analyzes the movies watched by a customer and attempts to “understand” what the customer likes
- 14. CAS-HC225 Spring 2024 14/17 ReinforcementLearning ● The model (“an agent”) with little or know prior knowledge “pokes” the environment ● The environment rewards the agent for “right” choices and penalizes for “wrong” choices (“carrot and stick”) ● Used in videogames, robotics, algorithmic trading, autonomous vehicles
- 15.
- 16.
- 17. CAS-HC225 Spring 2024 17/17 What’sNext? Neural Networks