Machine Learning Deep Learning AI and Data Science
9 likes9,214 views
The document provides a comprehensive overview of machine learning, deep learning, artificial intelligence, and data science, detailing definitions, applications, and interconnectedness of these fields. It discusses the significance of algorithms, data types, and various applications such as credit risk assessment and image recognition. Additionally, it outlines necessary skills and training paths for aspiring data scientists and dispels common myths associated with the profession.
![20
• Lot of experiments have shown that a deep network with less
parameters performs better than a shallow network
• For example deep network with hidden nodes [10,10,10,10] might
perform better than shallow network with [80] hidden nodes
• Deep neural networks are amazingly powerful.
• With sufficient number of hidden layers and nodes, we can fit a model
to any type of data
• They have the power to capture any amount of non linearity
DEEP NEURAL NETWORKS](https://image.slidesharecdn.com/machienlearningandai-190327090728/85/Machine-Learning-Deep-Learning-AI-and-Data-Science-20-320.jpg)
Machine Learning Deep Learning AI and Data Science
- 1. ML-MACHINE LEARNING DL-DEEP LEARNING AI – ARTIFICIAL INTELLIGENCE DS – DATA SCIENCE VENKATA REDDY KONASANI
- 2. PART-1 What is Machine Learning? 2
- 3. Activity Close you r eyes an d th in k ab ou t two terms – Mac h in e Learn in g an d A rtific ial Intelligen c e. • What is the first thing that comes to your mind when you hear these terms • Machine Learning • Artificial Intelligence 3
- 4. 4 WHAT AREYOUTHINKING ABOUT – ROBOTS?
- 5. 5 OR ANY AI BASED MOVIE?
- 6. HAVEYOUTHOUGHT ABOUT • Mathematics • Statistics • Datasets • Data Analysis • Optimization • Algorithms • Data Mining 6
- 7. Machine learning (ML) is the scientific study of algorithms and statistical models that computer systems use to effectively perform a specific task without using explicit instructions, relying on models and inference instead. 7 OK GOOGLE …WHAT IS MACHINE LEARNING?
- 8. WHAT REALLY IS MACHINE LEARNING ?-WIKIPEDIA
- 9. • Using historical data to make future predictions • Building models on historical data to predictions • Taking training data, building models on the training data using the models to make the future predictions • Making the machine learn the patterns in the data 9 IN SIMPLETERMS ..
- 10. 10 DATA IS IN DIFFERENT FORMS Numerical data Image data (pixel intensities) Video data (frames per second) Sound data (waves) Text data (tweets, comments, feedback)
- 11. APPLICATIONS OF NUMERICAL DATA 11 Identifying risky customers before offering a loan CREDIT RISK MODELS Do you receive any marketing calls? Have you ever received any marketing call for Audi car? MARKETING ANALYTICS Have you ever wondered, why only you are getting promotional offers on cloths and accessories where as I am getting offers on apartments? RETAIL SALES ANALYTICS How does a bank decide the potential fraud transactions from millions of credit card swipes? FRAUD ANALYTICS
- 12. • Face recognition – Using image as input data • Object recognition – Pixels is the input data • Digit recognition – Using text as image • Self Driving Cars – Using video data as input 12 APPLICATIONS OF MACHINE LEARNING – IMAGES ANDVIDEO DATA
- 13. IMAGE DATA IS ALSO NUMERICAL DATA Statinfer.com 13 Image data Human Vision Computer Vision -1 -1 -1 -1 -1 -1 -1 -1 0.9 -0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0.3 1 0.3 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -0 1 1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0.8 1 0.6 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0.5 1 0.8 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0.1 1 0.9 -0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -0 1 1 -0 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 0.9 1 0.3 -1 -1 -1 -1 0.5 1 0.9 0.1 -1 -1 -1 -1 0.3 1 0.9 -1 -1 -1 0.1 1 1 1 1 1 -1 -1 -1 -1 0.8 1 0.3 -1 -1 0.4 1 0.7 -0 -0 1 1 -1 -1 -1 -1 1 1 0.1 -1 0.1 1 0.3 -1 -1 -0 1 0.6 -1 -1 -1 -1 1 1 0.8 0.3 1 0.7 -1 -1 -1 0.5 1 0 -1 -1 -1 -1 0.8 1 1 1 1 0.5 0.2 0.8 0.8 1 0.9 -1 -1 -1 -1 -1 -0 0.8 1 1 1 1 1 1 1 1 0.1 -1 -1 -1 -1 -1 -1 -0 0.8 1 1 1 1 1 1 0.2 -1 -1 -1 -1 -1 -1 -1 -1 -1 -0 0.3 0.8 1 0.5 -0 -1 -1 -1 -1 -1
- 14. • Sentiment Analysis • Extraction of key topics in the data • Document Classification 14 APPLICATIONS ONTEXT DATA
- 15. PART-2 What is Deep Learning? 15
- 16. 16 • ANN- Artificial Neural Network • ANN is one of the technique in Machine Learning • ANN has input layer , hidden layer and output layer • For a really complex and non liner datasets we need several hidden layers • ANN with multiple hidden layers is known as deep neural network ANN
- 17. 17 • ANN with a single layer is known as shallow network • ANN with multiple hidden layers is known as deep neural network • Not just multiple hidden layers sometimes the type of hidden layer is also different. • This concept of solving problems with multiple hidden layers is known as deep learning DEEP LEARNING
- 18. DEEPVS SHALLOW NETWORKS 18 A neural network with single hidden layer is called a shallow network A neural network with more than one hidden layer is called deep neural network shallow network Deep network
- 19. DEEPVS SHALLOW NETWORKS 19 A single layer might not have the flexibility to capture all the non linear patterns in the data A deep network first learns the primitive features followed by high level features. This helps in building efficient models shallow network Deep network
- 20. 20 • Lot of experiments have shown that a deep network with less parameters performs better than a shallow network • For example deep network with hidden nodes [10,10,10,10] might perform better than shallow network with [80] hidden nodes • Deep neural networks are amazingly powerful. • With sufficient number of hidden layers and nodes, we can fit a model to any type of data • They have the power to capture any amount of non linearity DEEP NEURAL NETWORKS
- 21. DEEP LEARNING IS A SUBSET OF MACHINE LEARNING 21 Machine Learning Deep Learning
- 22. PART-3 What is Artificial Intelligence? 22
- 23. MACHINE LEARNING MODELS 23 Training data Build Model
- 24. MACHINE LEARNING MODELS 24 New data Apply Model Class1 Get Prediction This prediction can be right or wrong
- 25. MACHINE LEARNING MODELS 25 New data Apply Model Class2 Get Prediction One way models This prediction can be right or wrong
- 26. AI = MACHINE LEARNING MODELS + FEEDBACK LOOP 26 Training data Model
- 27. AI = MACHINE LEARNING MODELS + FEEDBACK LOOP 27 New data Apply Model Class2 Get Prediction Feedback Loop
- 28. AI = MACHINE LEARNING MODELS + FEEDBACK LOOP 28 Update Training data based on feedback Update the Model based on data Prediction Feedback Loop Class2
- 29. • Manual entry after going through test cases – Google maps • Indirect feedback collection based on user actions for - User click vs not click on your YouTube ad • Indirect feedback collection based on actions – In case of self driving car, hitting a wall is an action. 29 HOW IS FEEDBACK COLLECTED
- 30. • Self driving cars • SIRI / Ok-google • Alexa /Google home • Recommendation systems • Image recognition • Speech recognition • Spam filtering 30 APPLICATIONS OF AI
- 31. MACHINE LEARNING IS A SUBSET OF ARTIFICIAL INTELLIGENCE 31 Artificial Intelligence Machine Learning Deep Learning
- 32. PART-4 What is Data Science? 32
- 33. • Data Driven Decision making • Making sense out of data • Finding hidden patterns in the data • Analysis using not just machine learning models but also using data visualizations, intelligent reports • Most of the techniques and tools seen in data analysis in early days are now falling under data science 33 WHAT IS DATA SCIENCE?
- 34. • Mathematics • Statistics • Coding • Database management • Data Analytics • Predictive modelling • Machine Learning • Deep Learning 34 DATA SCIENCE IS A FUSION OF MANY FIELDS
- 35. DATA SCIENCE– FOUR MAJORTYPE OF SKILLS 35 Database Analytics & ML Bigdata Presentation
- 36. THETECHNIQUESYOU NEEDTO KNOW Database Knowledge •Data base Management •Data blending •Querying •Data manipulations •ETL Predictive Analytics & ML •Basic descriptive statistics •Advanced analytics •Predictive modeling •Machine Learning Big Data knowledge •Distributed Computing •Big Data analytics •Unstructured data analysis Presentation Skill •Data visualizations •Report design •Insights presentation 36
- 37. MACHINE LEARNINGTOOLS AND SOFTWARE'S Database tools SQL/MySql OLAP cubes Teradata DB2/Sql Server/ Oracle/ Informix/Exadata Analytical tools SAS/R/SPSS/Python Weka/MATLAB/ TensorFlow/OCR Big Data Tools Hadoop, Hive, Pig, Mahout, Spark, Java Presentation Tools Excel Tableau, Qlikview 37
- 38. DATA SCIENCE -DESIGNATIONS Database Developer ETL Developer MIS & DB Developer Data Architect Data Engineer Data Analyst Statisticians Business Analyst Data Scientist Bigdata Developer Hadoop Developer Software Engineer MIS Analyst Reporting Analyst Business Analyst 38
- 39. Data Science MACHINE LEARNING IS A PART OF DATA SCIENCE Statinfer.com 39 Statinfer.com Artificial Intelligence Machine Learning * These are individual interpretations Deep Learning
- 41. FAQ BY DATA SCIENCE ASPIRANTS • I want to be data scientist what training should I take? • I already have knowledge on few tools, what are my next steps? • What skill should I add to my profile to make it to next level? • I am new to data science, where can I start ? 41
- 42. You need training based on your skill level. Based on skill set we can divide the whole data science aspirants into four categories 1. Beginner - Completely new to Data Science and ML 2. Intermediate - MIS and Reporting Analyst 3. Advanced – Data Analyst and Predictive Modeler 4. Complete Data Scientist – ML, Hadoop, R, Python, DL, AI 42 CATEGORIES OF PROFILES
- 43. THE LEARNING PATH Statinfer.com 43 Tools & Coding R/SAS/Python/ Hadoop/Weka Basic Statistics and Mathematics Basic Algorithms - Regression, Classification and Segmentation Advanced ML Algorithms -Neural Networks, SVMs, Random Forest and Boosting Deep Learning Models CNN, RNN and LSTM AI Models Deep Q Learning Reinforced Learning Markov Decision process
- 44. THE LEARNING PATH – OUR SUGGESTIONS Statinfer.com 44 Tools & Coding R/SAS/Python/ Hadoop/Weka Basic Statistics and Mathematics Basic Algorithms - Regression, Classification and Segmentation Advanced ML Algorithms -Neural Networks, SVMs, Random Forest and Boosting Deep Learning Models CNN, RNN and LSTM AI Models Deep Q Learning Reinforced Learning Markov Decision process 1. Do not try to learn all the steps in one sitting. 2. You need to learn, absorb and then practise before you reach the next step Stage-1 Stage-2 Stage-3
- 45. THE LEARNING PATH – OUR SUGGESTIONS Statinfer.com 45 Tools & Coding R/SAS/Python/ Hadoop/Weka Basic Statistics and Mathematics Basic Algorithms - Regression, Classification and Segmentation Advanced ML Algorithms -Neural Networks, SVMs, Random Forest and Boosting Deep Learning Models CNN, RNN and LSTM AI Models Deep Q Learning Reinforced Learning Markov Decision process 1. R or Python. Both are really good. Pick any one of them 2. It also depends on your business problem 3. If you are planning to learn deep learning then go for python Stage-1 Stage-2 Stage-3
- 46. THE LEARNING PATH – OUR SUGGESTIONS Statinfer.com 46 Tools & Coding R/SAS/Python/ Hadoop/Weka Basic Statistics and Mathematics Basic Algorithms - Regression, Classification and Segmentation Advanced ML Algorithms -Neural Networks, SVMs, Random Forest and Boosting Deep Learning Models CNN, RNN and LSTM AI Models Deep Q Learning Reinforced Learning Markov Decision process 1. Do not start with stage-2 or stage-3 directly. 2. Strong fundamentals will make the learning easy in later stages. Stage-1 Stage-2 Stage-3
- 47. THE LEARNING PATH – OUR SUGGESTIONS Statinfer.com 47 Tools & Coding R/SAS/Python/ Hadoop/Weka Basic Statistics and Mathematics Basic Algorithms - Regression, Classification and Segmentation Advanced ML Algorithms -Neural Networks, SVMs, Random Forest and Boosting Deep Learning Models CNN, RNN and LSTM AI Models Deep Q Learning Reinforced Learning Markov Decision process 1. While learning these concepts, try to avoid academic style courses. 2. Look for the courses with lot of hands-on exercises and case studies Stage-1 Stage-2 Stage-3
- 48. THE LEARNING PATH – OUR SUGGESTIONS Statinfer.com 48 Tools & Coding R/SAS/Python/ Hadoop/Weka Basic Statistics and Mathematics Basic Algorithms - Regression, Classification and Segmentation Advanced ML Algorithms -Neural Networks, SVMs, Random Forest and Boosting Deep Learning Models CNN, RNN and LSTM AI Models Deep Q Learning Reinforced Learning Markov Decision process 1. Do not focus on the tool, focus on the technique and algorithm 2. Learning python or R tool, will not make you a data scientist Stage-1 Stage-2 Stage-3
- 50. FOCUS IS ON FIRSTTWO STAGES Statinfer.com 50 Tools & Coding R/SAS/Python/ Hadoop/Weka Basic Statistics and Mathematics Basic Algorithms - Regression, Classification and Segmentation Advanced ML Algorithms -Neural Networks, SVMs, Random Forest and Boosting Deep Learning Models CNN, RNN and LSTM AI Models Deep Q Learning Reinforced Learning Markov Decision process Stage-1 Stage-2 Stage-3
- 51. DURATION – 10 DAYS Statinfer.com 51 Tools & Coding R/SAS/Python/ Hadoop/Weka Basic Statistics and Mathematics Basic Algorithms - Regression, Classification and Segmentation Advanced ML Algorithms -Neural Networks, SVMs, Random Forest and Boosting Deep Learning Models CNN, RNN and LSTM AI Models Deep Q Learning Reinforced Learning Markov Decision process Stage-1 Stage-2 Stage-3
- 52. TWO PHASES Python for data science Data manipulations in python Basic Statistics Data validation and Cleaning Regression Logistic Regression Decision Trees Cluster Analysis Model Selection and Cross validation 52 PHASE-1 (5DAYS) ANN – Artificial Neural networks SVM – Support Vector Machines Random Forest Boosting NLP & Text mining TensorFlow & keras Deep Learning Models Convolution Neural Network Recurrent Neural Networks PHASE-2 (5DAYS)
- 53. • 100% Hands-on Training • 30 case studies laced in the course • Created for Non- Statisticians • Datasets from multiple domains, codes files and in class exercises • Team assignments and mentoring • Final Assessment • E-learning material support 53 COURSE FEATURES
- 54. PART-7 Data Science and Machine Learning Myths Statinfer.com
- 55. MYTH-1 : MATHEMATICS Myth-1 : To be a good data scientist, you need to be exceptional at statistics, mathematics, calculous, algorithms etc., Not necessarily. Statinfer.com 55
- 56. MYTH-2 : PROGRAMMING Myth-2 : To be a good data scientist, you need to have exceptional coding skills like Python, Java, C++ etc., Not necessarily. Statinfer.com 56
- 57. MYTH-3 : COMPLICATED MODELS Myth-3 :Data science is all about building complex predictive and machine learning models to solving business problems Not necessarily. Statinfer.com 57
- 58. MYTH-4 : MODEL BUILDING Myth-4 :After collection of the data, most of the time is spent on model building process. Not necessarily. Statinfer.com 58
- 59. MYTH-5 : LARGE DATASETS Myth-5:While solving machine learning problems we need to handle really large datasets or most of the datasets are really large Not necessarily. Statinfer.com 59
- 60. MYTH-6 : MACHINE LEARNING IN BUSINESS Myth-6:Companies use really advanced deep learning and AI models for while building all their business strategies Not necessarily. Statinfer.com 60
- 61. MYTH-7 : DIVERSE ALGORITHMS Myth-7: A data scientist will be using all the models in their day to day life Not necessarily. Statinfer.com 61
- 62. Have you tried our app? We have data science and machine learning quiz app to make the learning easy and fun Click here 62
- 63. 63 Our e-Learning Modules Currently all the Machine Learning and Data Science courses are available as e-learning modules
- 65. VENKATA REDDY KONASANI Trainer in • Data Science • Machine Learning • Deep Learning • Artificial Intelligence
- 66. ABOUTVENKATA REDDY KONASANI • 11 Years into Data Analytics • 5+ Years into Training • Author of the book “Practical Business Analytics using SAS” • Statinfer.com – Key member in the core team • Work Experience • HP – Data Scientist • Trend wise Analytics -Data Scientist • HSBC – Data Analyst • Citi – Risk Analyst • Masters in Applied Statistics and Informatics from IIT Bombay 66 5,000+ Training Hours 1,500+ Participants 75+ Corporate Batches 25+ Companies