Machine Learning an Exploratory Tool: Key Concepts
- 1. Prof. Amlan Chakrabarti IEEE Computer Soc. Dist. Vist. & ACM Dist. Speaker Director, A.K.Choudhury School of Information Technology University of Calcutta National Webinar On Education 4.0 “Ensuring Continuity in Learning and Innovation Through Digitization” Organized By: Singhad Institute of Management, Pune in Association with Savitribai Phule Pune University 12th June 2020
- 3. Formal Definition • “Machine learning is the field of study which gives the computers the ability to learn without being explicitly programmed”- Arther Samuels 1959 • “A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E.”- Tom Mitchells 1997 • Machine learning focuses on the development of computer programs that can access data and use it learn for themselves. • Example: To predict, traffic patterns at a busy intersection (task T) • We can run it through a machine learning algorithm with data about past traffic patterns (experience E) • If it has successfully “learned”, it will then do better at predicting future traffic patterns (performance measure P).
- 4. Defining The Learning Task
- 5. Why Machine Learning is Different?
- 6. Evolution of Machine Learning
- 7. Real World Problems lead to ML • The goal of ML is never to make “perfect” guesses, the goal is to make guesses that are good enough to be useful
- 8. Deep Learning in The Headlines
- 9. Deep Learning 9 Basics • A deep neural network consists of a hierarchy of layers, whereby each layer transforms the input data into more abstract representations (e.g. edge -> nose -> face) • The output layer combines those features to make predictions
- 10. Scene Labeling Using DL
- 11. Machine Learning for Smarter World
- 13. Machine Learning Techniques • Supervised Learning – Regression – Classification • Binary • Multiclass • Multi-label • Unsupervised Learning – Partitional – Hierarchical • Reinforcement • Semi-Supervised
- 15. Supervised Learning: Regression There are a few concepts to unpack here: • Dependent Variable • Independent Variable(s) • Slope & Intercept • Error Function
- 16. Supervised Learning: Classification Binary Classification Multi-label Classification Multiclass Classification
- 17. Unsupervised Learning • No labels are given to the learning algorithm, leaving it on its own to find structure in its input • The goal of unsupervised learning is to find hidden patterns in unlabeled data
- 18. Unsupervised Learning: Clustering • Finding groups of objects such that objects in a group are similar (or related) to one another and different from (or unrelated to) the objects in other groups Partitional Clustering Hierarchical Clustering
- 20. Reinforcement Learning • A technique to allow an agent to take actions and interact with an environment so as to maximize the total rewards • Similar to toddlers learning how to walk who adjust actions based on the outcomes they experience • A Playing Agent – Manages to score a point it gets a +1 reward – Each time it loses a point it gets a -1 penalty. – it will iteratively update its policies so that the actions that bring rewards are more probable and those resulting in a penalty are filtered out. • The first application in which reinforcement learning gained notoriety was when AlphaGo, a machine learning algorithm, won against one of the world’s best human players in the game Go
- 21. Semi Supervised Learning • Supervised Learning algorithm is a costly process, especially when dealing with large volumes of data • Unsupervised Learning is that it’s application spectrum is limited. • Semi-supervised learning combines a small amount of labeled data with a large amount of unlabeled data during training • Application Scenarios: Speech Analysis, Internet Content Classification, Protein Sequence Classification
- 22. Designing A Learning System • Choose the training experience • Choose exactly what is to be learned – i.e. the target function • Choose how to represent the target function • Choose a learning algorithm to infer the target function from the experience
- 23. NEURAL NETWORKS
- 24. Massively parallel interconnected network of simple processing elements which are intended to interact with the objects of the real world in the same way as biological systems do. NN models are extreme simplifications of human neural systems. Neural Networks
- 25. 25
- 26. 26
- 27. In context of Machine Learning, we want to learn the parameters from the training set, such that given the testing set data the training set can correctly classify the instances Artificial Neural Network (ANN)
- 28. ANN: Role of Weights & Bias Adjusting Weights Adjusting Bias
- 29. ANN: Understanding the Layers
- 30. ANN: Forward & Backward Propagation
- 31. The Big Picture
- 32. ML in Practice
- 33. ML Platforms Online Platforms Python Libraries
- 34. Future of Machine Learning • As ML assumes increased importance in business applications, this technology will be offered as a Cloud-based service known as Machine Learning-as-a-Service (MLaaS) • Connected AI systems will enable ML algorithms to “continuously learn,” based on newly emerging information on the internet • There will be a big rush among hardware vendors to enhance system power to accommodate ML data processing. More accurately, hardware vendors will be pushed to redesign their machines to do justice to the powers of ML • ML will help machines to work autonomously sense of context and meaning of data
Editor's Notes
- #14: Reinforcement: Learning through a reward mechnism Online: Doing something for every input you are getting Semi Supervised: Incorporates pseudo labelling Batch: Whole dataset at once