ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING
- 1. ARTIFICIAL INTELLIGENCE & MACHINE LEARNING Presentation by: Dr. SANDEEP RANJAN
- 2. TABLE OF CONTENTS • INTELLIGENCE • ARTIFICIAL INTELLEGENCE • ARTIFICIAL INTELLEGENCE SUBSETS • MACHINE LEARNING • APPLICATIONS OF MACHINE LEARNING
- 3. • INTELLIGENCE • Who is intelligent? • All living organisms are intelligent. • They interact with their environment and survive. • Examples from our own world ➢Crossing a road ➢Discovering alternate paths ➢Writing a poem, drawing a picture, creating a new recipe
- 4. • ARTIFICIAL INTELLIGENCE • Living beings are intelligent; but are man made non living beings also intelligent??? • Can a machine ➢make discoveries? ➢pass a ruling order in a court? ➢compose a symphony? ➢go for a PLAN B? ➢decide to wait or let go?
- 5. • ARTIFICIAL INTELLIGENCE • Traditional computers are powerful but not intelligent • They can compile MBs and GBs of code but may get stuck at a minor logical error • Artificial intelligence is a field of computer science which aims to make computer systems that can mimic human intelligence. • Just as we humans act when we don’t have exact information about a situation but still go ahead and choose one of the many possible moves.
- 6. • ARTIFICIAL INTELLIGENCE • Why make machines INTELLIGENT? • To reduce our effort and help the society advance ➢share our load ➢make use of massive number crunching power of CPUs ➢perceive things and try to realize them ➢perform in our absence/ without our guidance
- 7. • ARTIFICIAL INTELLIGENCE SUBSETS • MACHINE LEARNING • ARTIFICIAL NEURAL NETWORKS • DEEP LEARNING • COMPUER VISION • NATURAL LANGUAGE PROCESSING • SPEECH RECOGNITION
- 8. • MACHINE LEARNING • It is a branch of Artificial Intelligence that gives computers the capability to learn without being explicitly programmed. • Focus is on imparting “learning” to machines • Learning over time and iterations (similar to human experience) • No longer dependent on rule based programming • Real world data and observations are fed to the system
- 9. • MACHINE LEARNING • ML algorithms can be broadly categorized into ➢SUPERVISED ➢UNSUPERVISED ➢REINFORCED
- 10. • MACHINE LEARNING • SUPERVISED LEARNING • Uses ground truth and labeled data • Requires prior knowledge • Approximates the relationship between input and output • Mainly divided into CLASSIFICATION and REGRESSION • Naïve Bayes, Random Forest, Support Vector Machine, Neural Networks
- 11. • MACHINE LEARNING (SUPERVISED) • CLASSIFICATION • approximating a mapping function (f) from input variables (X) to discrete output variables (y) • Predicting a label • Spam/ non spam • Positive/ negative
- 12. • MACHINE LEARNING (SUPERVISED) • REGRESSION • Approximating a mapping function (f) from input variables (X) to a continuous output variable (y) • Predicting a quantity • Predict salary from age/experience data • Sales forecast
- 13. • MACHINE LEARNING • UNSUPERVISED LEARNING • No historical labels • Learn the inherent structure of data • Discover the trends in data • Mainly divided into CLUSTERING and ASSOCIATION
- 14. • MACHINE LEARNING (UNSUPERVISED) • CLUSTERING • Dividing the population into groups • Same group members resemble each other compared to other groups • Connectivity/ centroid/ distribution/density models • K Means, Hierarchical, KNN, PCA
- 15. • MACHINE LEARNING (UNSUPERVISED) • ASSOCIATION • Rule based learning model • Discover rules that describe large portions of your data • Product placement in malls • Eg people that buy X also tend to buy Y
- 16. • MACHINE LEARNING • REINFORCEMENT • Maximize reward in a given situation • Find the best possible behavior/ path • Input: initial state of the model • Output: many possible solutions to a given problem • Training: reward or punishment • Iterations: best solution is selected when reward is maximum
- 17. • MACHINE LEARNING TOOLS • PYTHON • MATLAB • R • KNIME • WEKA • PYTORCH • GEPHI • NODEXL • NETLOGO…………………………………
- 18. • APPLICATIONS OF MACHINE LEARNING • NATURAL LANGUAGE PROCESSING • SENTIMENT ANALYSIS • HANDWRITING RECOGNITION • SPEECH/ FACIAL RECOGNITION • CUSTOMER PROFILING (banks and financial institutions) • RECOMMENDATION SYSTEM (movie and e commerce) • CUSTOMER CHURN PREDICTION (telecom sector)