IEEE North Jersey Section

Policy optimization, which learns the policy of interest by maximizing the value function via large-scale optimization techniques, lies at the heart of modern reinforcement learning (RL). In addition to value maximization, other practical considerations arise commonly as well, including the need of encouraging exploration, and that of ensuring certain structural properties of the learned policy due to safety, resource and operational constraints. These considerations can often be accounted for by resorting to regularized RL, which augments the target value function with a structure-promoting regularization term, such as Shannon entropy, Tsallis entropy, and log-barrier functions. Focusing on an infinite-horizon discounted Markov decision process, this talk first shows that entropy-regularized natural policy gradient methods converge globally at a linear convergence that is near independent of the dimension of the state-action space, whereas the vanilla softmax policy gradient method may take an exponential time to converge. Next, a generalized policy mirror descent algorithm is proposed to accommodate a general class of convex regularizers beyond Shannon entropy, even when the regularizer lacks strong convexity and smoothness. Time permitting, we will discuss how these ideas can be leveraged to solve zero-sum Markov games. Our results accommodate a wide range of learning rates, and shed light upon the role of regularization in enabling fast convergence in RL.Co-sponsored by: North Jersey SectionSpeaker(s): Dr. Yuejie Chi, Agenda: Policy optimization, which learns the policy of interest by maximizing the value function via large-scale optimization techniques, lies at the heart of modern reinforcement learning (RL). In addition to value maximization, other practical considerations arise commonly as well, including the need of encouraging exploration, and that of ensuring certain structural properties of the learned policy due to safety, resource and operational constraints. These considerations can often be accounted for by resorting to regularized RL, which augments the target value function with a structure-promoting regularization term, such as Shannon entropy, Tsallis entropy, and log-barrier functions. Focusing on an infinite-horizon discounted Markov decision process, this talk first shows that entropy-regularized natural policy gradient methods converge globally at a linear convergence that is near independent of the dimension of the state-action space, whereas the vanilla softmax policy gradient method may take an exponential time to converge. Next, a generalized policy mirror descent algorithm is proposed to accommodate a general class of convex regularizers beyond Shannon entropy, even when the regularizer lacks strong convexity and smoothness. Time permitting, we will discuss how these ideas can be leveraged to solve zero-sum Markov games. Our results accommodate a wide range of learning rates, and shed light upon the role of regularization in enabling fast convergence in RL.Room: M105, Bldg: Muscarelle Center, M105, , 1000 River Road , Teaneck , New Jersey, United States, 07666, Virtual: https://events.vtools.ieee.org/m/304875

IEEE North Jersey Section offers "Big Data Marketing Research". Indeed.com lists 14,177 Data, Analyst, or Marketing jobs in the New York tri-state area daily! As an engineer, you never did marketing. Getting the MBA takes two years. This is the better alternative to learn marketing.This course deals with the collection, evaluation and analysis of the big data market-related information. Topics are: market research industry, problem definition, research process, focus group, secondary database, quantitative research, questionnaire design, sampling techniques, statistical testing, bivariate and multivariate correlation, communicating results and management reports. Using IBM SPSS software, you will perform detailed big data analysis in business, cyber security, criminal justice, green technology, healthcare, finance, social services, and marketing research firms.You will receive the IEEE completion certificate and earned 2.0 IEEE CEUs (converts to 20 PDHs). In addition, you may work as a data analyst or market researcher in any organization that needs your quantitative skills. Past attendees got jobs at Amazon, Bank America, Facebook, Google, IBM, Microsoft, Morgan Stanley, Oracle, Verizon, and other Fortune 500 firms.Instructor: Donald Hsu, Ph.D., has been a corporate manager and CEO for 27+ years and is an experienced trainer. Since 2017, he has trained 500+ people in Management, Marketing, Global Marketing, and Marketing Research courses in seven organizations. He does international business in 90 countries.Co-sponsored by: New Jersey Institute Technology, CAS/ED, AP/MTTAgenda: - Describe the big data market research industry, problems and research process- Understand the importance of primary big data collection, secondary database, and survey- Define quantitative research, measurement technique and sampling methods for big data- Explain the questionnaire design, big data processing and statistical testing- Build the knowledge of bivariate regression and multivariate data analysis- Communicate results, manage ethical issues, and prepare reports- Employ IBM SPSS software for frequency analysis, ANOVA, T-test and others- Review real-world marketing research using Harvard Business School cases- Present final Group ProjectWHERE:Zoom or New Jersey Institute Technology, Newark, New JerseyWHEN:7 Saturdays, March 26, April 2, 9, 16, 23, 30, May 7, 2022 @9 am - 12 pmCOST:IEEE (& affiliate) members $500; Non-IEEE members $550.CONTACT:Donald Hsu: [email protected]: Big Data Market ResearchPlease mail the registration form with the check (Checks payable to “North Jersey Section IEEE”) toDr. Donald Hsu, Chair Education Committee, IEEE North Jersey Section, P. O. Box 2093, Fort Lee, New Jersey 07024Name: _____________________________________________ Email address _________________________________ÿ Non-memberÿ IEEE Member Member #:_________________________ Member of _____________________________ technical societyEmployer:_______________________ Employer Address:__________________________________________________________Home Address:___________________________________Business (day) telephone #:___________________________________ Home telephone #:________________________________Please enclose required fee payable to: North Jersey Section IEEEI wish to receive the IEEE Completion Certificate Signature:___________________________________________Warren Street, Newark, New Jersey, United States