Machine Learning Methods 2.pptx
- 2. Machine Learning • ML is the process by which a computer learns from experience (e.g., using programs that can learn from historical cases)
- 3. Supervised Learning • The majority of practical machine learning uses supervised learning. • Supervised learning is where you have input variables (x) and an output variable (Y) and you use an algorithm to learn the mapping function from the input to the output. • Y = f(x) • The goal is to approximate the mapping function so well that when you have new input data (x) that you can predict the output variables (Y) for that data.
- 4. • It is called supervised learning because the process of an algorithm learning from the training dataset can be thought of as a teacher supervising the learning process. We know the correct answers, the algorithm iteratively makes predictions on the training data and is corrected by the teacher. Learning stops when the algorithm achieves an acceptable level of performance.
- 5. • Supervised learning problems can be further grouped into regression and classification problems. – Classification: A classification problem is when the output variable is a category, such as “red” or “blue” or “disease” and “no disease”. – Regression: A regression problem is when the output variable is a real value, such as “dollars” or “weight”.
- 6. • Some common types of problems built on top of classification and regression include recommendation and time series prediction respectively. • Some popular examples of supervised machine learning algorithms are: – Linear regression for regression problems. – Random forest for classification and regression problems. – Support vector machines for classification problems.
- 7. Unsupervised Machine Learning • Unsupervised learning is where you only have input data (X) and no corresponding output variables. • The goal for unsupervised learning is to model the underlying structure or distribution in the data in order to learn more about the data. • These are called unsupervised learning because unlike supervised learning above there is no correct answers and there is no teacher. Algorithms are left to their own devises to discover and present the interesting structure in the data.
- 8. • Unsupervised learning problems can be further grouped into clustering and association problems. – Clustering: A clustering problem is where you want to discover the inherent groupings in the data, such as grouping customers by purchasing behavior. – Association: An association rule learning problem is where you want to discover rules that describe large portions of your data, such as people that buy X also tend to buy Y.
- 9. • Some popular examples of unsupervised learning algorithms are: • k-means for clustering problems. – The objective of k-means is simple: group similar data points together and discover underlying patterns. To achieve this objective, k-means looks for a fixed number (k) of clusters in a dataset.” • Apriori algorithm for association rule learning problems. – Apriori algorithm is a classical algorithm in data mining. It is used for mining frequent item sets and relevant association rules
- 10. Case-Based Reasoning (CBR) • A case has two parts: a problem and a solution • Cases represent experience; that is, they record how a problem was solved in the past • CBR is a methodology in which knowledge and/or inferences are derived from historical cases. It is based on the premise that new problems are often similar to previously encountered problems and that, past solutions may be of use in the current situations. • CBR is particularly applicable to problems in which the domain is not understood well enough for a robust statistical model or system of equations to be formulated.
- 11. Process of CBR 1. Retrieve Given a target problem, retrieve the most similar cases 2. Reuse Map the solution and reuse the best old solution to solve the current case 3. Revise Test the solution and, if necessary, revise the old case to come up with the solution 4. Retain After the solution has been successfully adapted to the target problem, store the resulting experience as a new case
- 12. Step-by-Step Process of CBR
- 13. Similarity Computation • Cases are ranked according to their similarity based on the similarity of each feature • The degree of similarity can be expressed by a real number between 0 (not similar) and 1 (identical). • The importance of different features may be different. In that case, similarity is computed by weighted average.
- 14. CBR Examples • Intelligent customer support and sales support • Retrieval of tour packages from travel catalogs • Conflict resolution in air traffic control • Conceptual building design aid • Conceptual design aid for electronic devices • Medical diagnosis • Aircraft troubleshooting • Heuristic retrieval of legal knowledge • Computer supported conflict resolution through negotiation or mediation
- 15. Advantages and Disadvantages of Using CBR • Advantages – Improved knowledge acquisition – Reduced development time – Easier explanation – Learning over time • Disadvantages – Storing of cases in the KB. – Implicit link between problem and solution – Access and retrieval speed
- 16. GENETIC ALGORITHMS • G.As are programs that attempt to find optimal solutions to problems by conceptually following steps inspired by the biological processes of evolution • The method learns by producing offspring that are better and better, as measured by a fitness-to-survive function, until an optimal or near - optimal solution is obtained.
- 17. G. A. Fundamentals • Chromosome – A candidate solution for a genetic algorithm • Fitness function – A measure of the objective to be obtained. • Generation – An iteration of the genetic algorithmic process in which candidate solutions are combined to produce offspring
- 18. Processes within Genetic Algorithm • Reproduction – Through reproduction, genetic algorithms produce new generations of improved solutions by selecting parents with higher fitness ratings or by giving such parents a greater probability of being contributors and by using random selection. • Crossover – The combining of parts of two superior solutions by a genetic algorithm in an attempt to produce an even better solution • Mutation – A genetic operator that causes a random change in a potential solution
- 20. Genetic Algorithm Parameters • Some parameters must be set for the genetic algorithm – Number of initial solutions to generate – Number of offspring to generate – Number of parents and offspring to keep for the next generation – Mutation probability – Probability distribution of crossover point occurrence • Their values are dependent on the problem being solved and are usually determined through trial and error
- 21. Genetic Algorithm Benefits and Limitations • Genetic algorithms are particularly useful for complex problems that require rapid development of set of good solutions • Limitations – Not all problems can be framed in the mathematical manner that genetic algorithms demand • Development of a genetic algorithm is complex • In some situations, the “genes” from a few comparatively highly fit (but not optimal) individuals may come to dominate the population, causing it to converge on a local maximum • Most genetic algorithms rely on random number generators that produce different results each time the model runs
- 22. Genetic Algorithm Applications • Genetic algorithms provide a set of efficient, domain-independent search heuristics for a broad spectrum of applications including: – Dynamic process control – Complex design of engineering structures – Scheduling – Transportation and routing – Layout and circuit design – Telecommunications – Discovery of new connectivity typologies
Editor's Notes
- #15: Aircraft Trouble shooting is an ideal tool to gauge the accuracy with which the aircraft is flying a glideslope and can be used to cross check against other information. The FPV is an ideal tool to monitor non-automation phases of the flight (manual flying) as the flight crew need only to keep the FPV on the horizon to maintain
- #23: Scheduling Algorithms. To decide which process to execute first and which process to execute last to achieve maximum CPU utilisation, computer scientists have defined some algorithms, they are: First Come First Serve(FCFS) Scheduling. Shortest-Job-First(SJF) Scheduling