ML - Chapter 2 - Types of Machine Learning PDF

Types of Machine Learning and ML projects’ Lifecycle Presented by: Pr. Asmae BENTALEB Année Universitaire 2024-2025  Plan  What is Machine Learning?  Types of Machine Learning  Supervised  Unsupervised  Reinforcement learning  ML project lifecycle  The Definition...

Types of Machine Learning and ML projects’ Lifecycle Presented by: Pr. Asmae BENTALEB Année Universitaire 2024-2025  Plan  What is Machine Learning?  Types of Machine Learning  Supervised  Unsupervised  Reinforcement learning  ML project lifecycle  The Definition of Machine Learning The field of study that gives computers the ability to learn without being explicitly programmed (Arthur Samuel). The Machine Learning program performed better than Arthur himself. Supervised Learning Supervised Learning X  Y input output  Supervised Learning  The objective : to find the mapping between X and Y, given a certain dataset.  During the training, the algorithm is given the input X and label Y, so that later, when given a new X, the model can map it to a new Y.  Learning a function mapping X to Y.  It learns from being given the right answer, and by experience. Supervised Learning (examples of use cases) Input (X) Output(Y) Application Phone image Defect(yes/no) Quality control of phones Image of radar Position of car Auto driven cars email Spam(0/1) Spam filtering Voice record text Speech recognition  Supervised Learning: Regression  Predict a continuous value given inputs X : Example: predicting the price of a house/Car  Supervised Learning: Classification  Classification is used to predict categories: (a precised number of outputs, for example A and B)  Supervised Learning:  It learns from being given the right answer:  Regression: Predict a number infinitely many possible outputs  Classification: predict categories small number of possible outputs Unsupervised Learning  UnSupervised Learning Find something interesting in unlabeled data. Data only comes with inputs x, but no output labels y. Algorithm has to find a structure in the data. For example: the figure refers to Cluster analysis (clustering) which organizes unlabelled data points into groups according to similarity (or dissimilarity).  UnSupervised Learning It is the concept of using unlabled data, Given the features X and finding interesting Structures or patterns in the data.  UnSupervised Learning (example) Having many people talking, using unsupervised learning, voices can be separated. Here, we have no labels, and we are trying to separate the people’s voices that are overlapping as they were talking at the same time. The used algorithm is :independent component analysis UnSupervised Learning (example) Reinforcement Learning  Reinforcement Learning overview Reinforcement learning is a feedback-based learning method, in which a learning agent gets a reward for each right action and gets a penalty for each wrong action. The agent learns automatically with these feedbacks and improves its performance. In reinforcement learning, the agent interacts with the environment and explores it. The goal of an agent is to get the most reward points, and hence, improve its performance. reinforcement learning continuously improves its model by leveraging feedback from previous iterations. Example: Teaching a dog robot to behave, reward on good behavior and punish on bad one, so that with time, it will learn. Strategic decisions when building ML projects Collecting more data?! What types of data? What ML algorithm to choose? regression, classification..? Do we need more GPUs? Life cycle of a Machine Learning project Fig : Machine Learning (ML) Model Development Lifecyle