Post on 10-Jan-2017
APPLIED INNOVATIONS IN
MACHINE LEARNING AND
SOFT COMPUTING THE USAABHISHEK SYAL
February, 2016NITTTR, CHANDIGARH
ABHISHEK SYAL
About me■ MBA from MIT Sloan, B.E. (Hons.) BITS,
Pilani ■ Market Intelligence at EMC Corp., Research
at BHEL Corp. R&D using Predictive Analytics
■ Co-inventor to 4 patent pending tech
■ Founded and led my own social venture, ARISE– Differentiated offering: self-learning
for disabled for over 300+ children; proprietary methodology
Agenda■Part – I: Brief Introduction to Machine Learning
and Soft Computing
■Part – II: Applied Innovations in the USA
■Part – III: Research Areas
BRIEF INTRODUCTION TO MACHINE LEARNING AND SOFT COMPUTING
PART - I
Machine Learning■Machine learning is the science of getting computers to
act and respond without being explicitly programmed
■Type of AI where computers learn and grow better with time to solve for the objective function
■Data analysis and actions using automatic model building, many of them using mathematical optimization techniques
Selected Machine Learning Techniques■ANN: Artificial Neural Network – group of artificial
neurons
■Clustering: putting a set of object into clusters, building compactness
■Decision Tree Learning: Using decision trees as predictive models
Soft Computing
■Use of inexact solutions to computationally hard tasks, difficult to solve for in the time objective
■Tolerant for imprecision, uncertainty, partial truth and approximation
■Role model for soft computing is human mind
Selected Soft Computing Techniques■Artificial Neural Networks
■Fuzzy logic: many valued logic between 0 and 1, usage of linguistic variables
■Bayesian Network: probabilistic graphical model with set of random variables and their conditional dependencies in a Directed acyclic graph (directed graph with no directed cycles)
APPLIED INNOVATIONS IN THE
USAPART - II
1. Energy Management and Efficiency■My experience is in designing a meta-heuristics
controller application using model heuristics for Operational Efficiency
■Cloud computing and IIoT (Industrial Internet of Things) to aid in real-time grid optimization in terms of connectivity, load balancing and remote operation
■Preventive Maintenance of energy sources and distribution infrastructure, saving CapEx and OpEx
■Home Automation and Building Energy Management
GridPoint is an innovator in comprehensive, data-driven energy management solutions (EMS) that leverage the power of real-time data collection, big data analytics and cloud computing to maximize energy savings, operational efficiency, capital utilization and sustainability benefits.
2. Human Machine Interaction■Speech recognition (e.g. Alexa)■Web search ■Face and gesture recognition ■Image Identification and classification ■Applications in mobile retail, social media, security,
fraud detection and surveillance
3. Infrastructure and Transport■ Navigation■ On-demand match of supply and demand for uber,
olacabs, etc. – GPS as the main sensor, supply agents (cabs), inputs
(ask requests)■ Self-driving cars■ Drone flights■ Construction & Farms
– Inspection– Survey
Other Applications■ Understanding human genome■ Fraud detection
– Credit card– Internet
■ Brain-machine interfaces■ Computational finance■ Sentiment analysis ■ Online Advertising■ Robotics■ Augmented Reality
RESEARCH AREASPART - III
Research Areas
■Efficiency management, preventive maintenance and optimization of dynamic multi-agent systems
■Humanization of machine-human interactions for more life-like natural experiences as well as to prevent fraud
■Hyper-personalization, localization and contextualization of automation systems with added learning
THANKSabhishek.syal@sloan.mit.edu