1989 — 1993 |
Glass, Arnold |
N/AActivity Code Description: No activity code was retrieved: click on the grant title for more information |
Syntactic Processing and Grammatical Intuition @ Rutgers University New Brunswick
This research derives from the notion that the process by which a sentence is comprehended requires several steps. One step is to retrieve the meanings of the words, and another step is to combine the meanings of the individual words into the meaning of the entire sentence. The mental procedure that performs these tasks is called the "parser", because it "parses" sentences. How the meanings of the words are combined is dictated by their inflections and by their order in the sentence. For example the sentence "Boys like girls." sounds natural because the words and their inflections form a pattern that induces the parser to combine the meanings of the words in such a way that we understand "boys" to be the agent and "girls" to be the recepient of the action "like." However, not all patterns of words and inflections are recognized by the parser. In this view, the more dissimilar a pattern is from one recognized by the parser, the more odd or deviant the string of words sounds. Hence, strings like "Boy like girl." and "Boys red girl." sound deviant. Such strings are ungrammatical; the process by which people judge whether a string in the language is natural or odd is known as grammatical intuition. Several predictions may be derived from this notion about the role of grammatical intuition in language comprehension. First, it should be possible to specify what the patterns are that the parser recognizes and to show that the perceived deviance of a word string increases with its difference from the pattern most similar to it that the parser recognizes. Also, deviant word strings should be more difficult to understand than natural ones. This research project will test these predictions. First, the parsing procedure for a subset of English will be implemented in a computer program. The specificity of the program will make it possible to derive precise predictions about grammaticality and comprehensibility for the sentences recognized by the program. Then, these predictions will be compared with measures of human performance; among them, the time college students take to make grammaticality judgments for the sentences and the time they take to comprehend the sentences. Regardless of the merits of the initial hypothesis, comparison of the performance of a computer program with human performance is likely to produce results that will improve our understanding of human language comprehension and hence that will be useful in improving the ability of computer programs to understand human language.
|
0.915 |
2014 — 2017 |
Glass, Arnold Mandayam, Narayan [⬀] |
N/AActivity Code Description: No activity code was retrieved: click on the grant title for more information |
Nets: Small: End-User Behavior and Prospect Pricing in Wireless Data Networks @ Rutgers University New Brunswick
Beyond smart phones, of which there are significantly more in developed countries, the number of mobile phones in the world is approaching 7B, even creating a reality that in some parts of the world, there are more people with access to a phone than with access to electricity at home. The advent of machine-to-machine communications adds increased pressure on wireless system capacity. In fact, there is a recognition and push in both industry and academia towards the goal of achieving "1000x" capacity for wireless. The solution approaches range from spectrally agile cognitive radios with novel spectrum sharing, to use of higher frequency electromagnetic spectrum as well as smaller and denser cell deployments referred to as heterogeneous networks (HetNets). While this is a much needed activity with many challenges to overcome, providing a spatially high density of wireless/wired backhaul as required for HetNets is expensive and the overwhelming demands on wireless capacity fundamentally remain, in that state-of-the-art systems are nowhere near the 1000x capacity target goals and perhaps even an order of magnitude or two away. Wireless service providers (SPs) in recent times have therefore resorted to control access and services being provided to end-users via differentiated and hierarchical monetary pricing. As such, end-users may have to make decisions on data rate and price offerings that may be presented to them when they need service in high user-density dynamic spectrum settings. A complementary approach termed "prospect pricing" is proposed as a way to support data demand and relies on influencing end-user (human) behavior using dynamic pricing algorithms when technological solutions by themselves cannot satisfy the demands of wireless data. The research agenda seeks to design and study wireless network pricing from a cognitive psychology perspective, thereby presenting a novel framework to understand how wireless networks can be influenced by end-user behavior and vice-versa. The successful completion of this research will serve up useful pointers to how prospect pricing can be used by the SPs to manage the ever increasing demand for data.
Policing mechanisms that influence wireless device behavior and thereby drive systems to better operating points have been addressed amply in the radio resource management literature. These mechanisms essentially are borne out of expected utility theory (EUT) based microeconomics approaches, and implemented via engineered system design, i.e., embedding these strategies in the link layer and network layer protocols that are executed by wireless devices. When a SP controls access to end-users via differentiated and hierarchical monetary pricing, then the performance of the network is directly subject to end-user decision-making that has shown to deviate from EUT. Prospect Theory, a Nobel prize winning theory that explains real-life decision-making and its deviations from EUT behavior, is used to design "prospect pricing" for wireless networks. Specifically, dynamic pricing algorithms for wireless data are designed to enable HetNets to manage the ever increasing demand for data, especially when both spectrum and infrastructure resources are constrained. Using a mix of theory, algorithm development and experimentation, the research agenda proposed by a team comprised of a wireless networking/systems researcher and a cognitive psychologist includes: (1) Development of a Framework for Prospect Pricing in Wireless Networks using Game Theory, (2) Evaluation of the Performance of Prospect Pricing in HetNets for Load-Balancing and Resource Management, and (3) Psychophysics Experiments to understand End-User Perceptions and Preferences to Service Offers and Wireless Network Performance.
The project will make available as an open resource the psychophysics testbed developed for wireless network usage experience. The unique marriage of wireless network pricing and cognitive psychology offers an innovative educational opportunity to involve both graduate and undergraduate students from electrical and computer engineering and psychology.
|
0.915 |
2015 — 2017 |
Mandayam, Narayan [⬀] Glass, Arnold |
N/AActivity Code Description: No activity code was retrieved: click on the grant title for more information |
Eager: Renewables: Collaborative Research: Foundations of Prosumer-Centric Grid Energy Management @ Rutgers University New Brunswick
This project considers the interaction of prosumers with the power grid using prospect theory, a seminal contribution to behavioral economics that won the Nobel Prize (?prosumer? refers to an electricity consumer who also takes the role of a producer and seller of electricity). Prospect theory gives a methodology for understanding people?s economic choices based on their actual behavior and their assessment of potential gains and losses versus the assessed levels of risk. Prospect theory thus helps one to understand the actual dynamics of economic choices, without assuming that the outcomes that might be predicted using game theory must actually come to pass. The project takes the stand that achieving a sustainable smart grid requires transforming today?s power grid into a prosumer-centric grid in which a significant portion of energy stems from renewable sources and other prosumer-owned devices. The success of this vision is contingent on large-scale, active prosumer participation in energy management. However, just because such participation can be of significant benefit to participants, it cannot be assumed that prosumers will actually become fully involved in the smart grid. Empirical data shows that, despite its exciting prospects, the widespread adoption of the smart grid has been hindered by modest user participation. Motivated by emerging grid scenarios, this project employs the mathematical framework of prospect theory to study smart grid energy management. The results of the project will advance multiple disciplines including power systems, game theory, economics, and cognitive psychology. The behavioral experiments will pioneer the generation of new real life models for user participation in energy management. This interdisciplinary research will provide necessary mathematical foundations to expedite the realization of a prosumer-centric, sustainable smart grid vision. The developed techniques will provide a fundamental understanding of the role of prosumers in the smart grid. It can also lead to new economic and regulatory grid protocols and policies. A software tool will create a new platform for participation in smart grid research and education. The unique marriage of smart grid design and cognitive psychology will offer an innovative educational opportunity to involve graduate and undergraduate students from both engineering and psychology via new and existing courses as well as participation in behavioral experiments. Outreach events targeted at high school and minority students will be organized.
There is ample evidence (anecdotal and otherwise) that decision making in real life is often guided by perceptions that deviate from the precepts of standard game theory. Using prospect theory to understand the role of such users? perceptions in the smart grid, an increasingly prosumer-centric system, is a stepping stone toward accelerating its adoption. Preliminary investigation of prospect theory in energy management reveals that deviations from standard game theory can lead to undesirable grid performance, thus necessitating new prospect theory schemes cognizant of real life prosumer behavior. Using prospect theory, a novel framework is introduced to fundamentally understand the role of prosumer participation in energy management. Due to an interdisciplinary mix of theory, algorithm design, and cognitive psychology experiments, this research will yield the following key innovations: 1) new fundamental results on the impact of prosumer behavior on energy management, 2) prosumer-centric, sustainable energy management schemes and associated pricing mechanisms that optimize grid operation by tightly integrating the effect of user behavior and subjective utility perceptions via novel prospect theory notions, 3) a new stochastic prospect theory game class for grid-aware energy management that accounts for grid dynamics and uncertainty due to factors such as renewables, and 4) real life cognitive psychology experiments coupled with a new software platform that will yield new, realistic behavioral models for smart grid prosumers.
|
0.915 |
2016 — 2019 |
Lindqvist, Janne Mandayam, Narayan [⬀] Glass, Arnold |
N/AActivity Code Description: No activity code was retrieved: click on the grant title for more information |
Crisp Type 2: Collaborative Research: Towards Resilient Smart Cities @ Rutgers University New Brunswick
Realizing the vision of truly smart cities is one of the most pressing technical challenges of the coming decade. The success of this vision requires synergistic integration of cyber-physical critical infrastructures (CIs) such as smart transportation, wireless systems, water networks, and power grids into a unified smart city. Such smart city CIs have significant resource dependence as they share energy, computation, wireless spectrum, users and personnel, and economic investments, and as such are prone to correlated failures due to day-to-day operations, natural disasters, or malicious attacks. Protecting tomorrow's smart cities from such failures requires instilling resiliency into the processes that manage the city's common CI resources. Such processes must be able to adaptively and optimally reallocate smart city resources to recover from failure. The goal of this Critical Resilient Interdependent Infrastructure Systems and Processes (CRISP) collaborative research project is to address this fundamental challenge via a coordinated and interdisciplinary approach that relies on machine learning, operations research, behavioral economics, and cognitive psychology to lay the mathematical foundations of resilient smart cities. The anticipated results will break new ground in the understanding of synergies between multiple cyber-physical infrastructure and resilient resource management thus catalyzing the global deployment of smart cities. This research will yield advances to the areas of resilient systems, cyber-physical systems, security and privacy engineering, game theory, computer and network science, behavioral economics, data analytics, and psychology. The project will involve students from diverse backgrounds across engineering, computer science, economics, and psychology that will be trained on pertinent research issues related to smart cities and resiliency. The project will also contribute to fostering trust between residents and the various technological processes that are fundamental to the operation of a smart city.
This research will introduce a foundational, transformational, analytical framework for leveraging synergies between a city's CIs to yield resilient resource management schemes cognizant of both technological and human factors. By bringing together researchers from interdisciplinary fields, this framework yields several advances: 1) Rigorous mathematical tools for delineating the inter-dependencies between CIs via a complementary mix of novel tools from graph theory, power indices, machine learning, and random spatial models; 2) Resilient resource management mechanisms that advance notions from frameworks such as behavioral game theory to enable optimized management of shared CI resources in face of failures stemming from agents of varying intelligence levels; 3) Behavioral models for characterizing the trust relationships between the residents of a smart city and the CIs; 4) Behavioral studies that provides guidelines on how to influence the users of the CIs in such a way so as to improve the resiliency of the CIs; and 5) Large-scale smart city simulators coupled with realistic experiments that will bridge the gap between theory and practice. The insights from this project will apply to the future scientific cyber-infrastructures that are likely to be interconnected as well as interdependent. The simulator will be a software artifact that would be a useful component of a scientific cyberinfrastructure aimed at understanding (for example) smart cities.
|
0.915 |