Research & Projects
Research & Projects
Embedded Files
Strategic quantization
Strategic quantization
Quantization setting with misaligned objectives for the encoder and the decoder. While the unconstrained variation of this problem has been well-studied under the theme of information design problems in Economics, the problem becomes more appealing and relevant to engineering applications with a constraint on the cardinality of the message space. We show that a straightforward extension of Lloyd-Max algorithm for strategic scenarios may not converge or may converge to a solution that is not locally optimal. We develop a gradeint descent based algorithm to mitigate this issue. However, we show that even for log concave sources there may be multiple local optima, unlike the case for classical quantization. We extend dynamic programming solutions in classical quantization literature to our setting for global optimality.
isit_2023.pdfdcc_2022.pdfChannel-optimized strategic quantization
Channel-optimized strategic quantization
We consider the strategic quantization setting described above with communication over a noisy channel. The solution method uses gradient descent/dynamic programming with random index mapping.
Anand and Akyol. “Strategic Quantization over a Noisy Channel”, ASILOMAR’23.
Anand and Akyol, “Channel-Optimized Strategic Quantizer Design via Dynamic Programming”, SSP'23.
asilomar_2023.pdfssp_2023.pdfStrategic quantization of a noisy source
Strategic quantization of a noisy source
We explore the problem of strategic quantization of a remote source observed through a noisy channel. We show that as in the classical noisy source setting, the indirect source coding in the strategic setting can also be transformed to a direct source coding with a modification in the distortion measure that is the original distortion is conditioned on the sensing channel output.
Anand and Akyol. “Strategic Quantization of a Noisy Source”, ALLERTON’23.
allerton_2023.pdfImpact of bounded rationality in strategic data gathering
Impact of bounded rationality in strategic data gathering
We study the problem of debiasing survey data gathered from strategic and boundedly rational agents with heterogenous objectives and partial information. Specifically, our setting involves K different types of survey responders with varying levels of available information, degree of strategic behavior, and cognitive hierarchy: i) a non-strategic agent with an honest response, ii) a k−th level (k = 1, 2, ..., K-1) strategic agent that believes the population is Poisson distributed over the lower cognitive types.
Anand and Akyol. “On the Impact of Bounded Rationality in Strategic Data Gathering”, accepted in CPHS'24.
cphs-2024.pdfIncorporating fairness into electricity pricing games
Incorporating fairness into electricity pricing games
This paper investigates the incorporation of fairness considerations, as defined in the behavioral economics literature, into game-theoretic formulations of dynamic pricing algorithms in electricity markets. Traditionally, these markets are modeled using Stackelberg equilibria between electricity producers (companies) and consumers, where the producer acts as the leader by setting the electricity price, and consumers respond by adjusting their load. This conventional order of play inherently limits the producer’s ability to arbitrarily increase prices, as the load is price-dependent and the producer commits to supplying a minimum load prior to the interaction. Recent analyses, however, indicate that the order of play may reverse in practical scenarios where players converge to equilibrium through learning dynamics. In this reversed hierarchy, fairness constraints become pivotal in preventing the producer from opportunistically escalating prices without bound.
Roberts, Anand and Akyol. “Incorporating Fairness into Electricity Pricing Games”, accepted in ICC'25 .
icc_2024.pdfLink to codes: strategic-quantization (github.com), fair-dynamic-pricing
Page updated
Report abuse