Machine learning presentation
Machine learning is motivated by the ability to use data to build models that can make predictions without being explicitly programmed. There are three main types of machine learning: supervised learning where labeled data is used to train models, unsupervised learning where unlabeled data is used to find hidden patterns in data, and reinforcement learning where agents learn from rewards and punishments. The machine learning process involves getting data, preparing it, training models on the data, testing the models, and improving the models through iterative training. Popular tools for machine learning include Python programming languages, TensorFlow framework, cloud platforms like Google Cloud, and resources like online courses and Kaggle competitions.
Machine learning presentation
- 2. Agenda Motivation behind ML Concept of the ML Technique of the ML 2
- 3. Motivation Whats Machine learning ? Where using Machine learning ? The future of Machine learning ? 33
- 4. Part 1 : Concept of the ML 44
- 5. Concept Sequential logic and Machine learning Types of Algorithms 5 5
- 6. Sequential logic and Machine learning Sequential logic Machine learning Input Output label Model 6 6
- 7. Type of learning Supervised Unsupervised Reinforced 7 7
- 8. Supervised Input data are already labeled Model train by iterating Commun exemples : 1) Regression (predicting value) 8 8
- 9. 2) Classification (predicting Label) Supervised 9 9
- 10. Unsupervised Input data are not labeled Model deducer structures Commun exemples : 1) Clustering 10 10
- 12. 12 Reinforcement Learn by using a reward function Commun exemples : 1) Games 1212
- 13. Part 2 : Technique of the ML 13 13
- 14. Technique The machine learning process Real example Share tools for you Touch and play with some tools (Practical) 14 14
- 15. 1515
- 16. Get Data Need lots of data Lots of excellent ones are online 16 16
- 17. Clean , prepare data Convert into appropriate format (using python) Label the data Distort data to generalize you model Make a good model first to avoid problems 17 17
- 18. Train model Very computationally expensive(GPU,CPU,TPU) Only experts can configure from scratch (using cloud) 18 18
- 19. Testing Typically saving some train data for testing Test for worst state for highest purpose 19 19
- 20. Improve Iterative and increase accuracy 20 20
- 22. Tools for you Programming languages : R , Python (recommended ) Platform : Any typing editor Framework : Tenserflow lib Clouds : floydhub (free to test) Google cloud , amazon or azure (train model) Maths : Linear algebra , Calculus , probability and statistic Library and Competition : https://www.kaggle.com/ Lecture (courses) : Coursera , Udacity Books : building machine learning systems with python (for beginners) 22 22
- 23. Bye 2323