International Society of Dynamic Games

Dynamic Games and Applications Seminar

ShapG: new feature importance method based on the Shapley value

Jan 23, 2025   11:00 AM — 12:00 PM (Montreal time)

Zoom webinar link

With wide application of Artificial Intelligence (AI), it has become particularly important to make decisions of AI systems explainable and transparent. In this paper, we proposed a new Explainable Artificial Intelligence (XAI) method called ShapG (Explanations based on Shapley value for Graphs) for measuring feature importance. ShapG is a model-agnostic global explanation method. At the first stage, it defines an undirected graph based on the dataset, where nodes represent features and edges are added based on calculation of correlation coefficients between features. At the second stage, it calculates an approximated Shapley value by sampling the data taking into account this graph structure. The sampling approach of ShapG allows to calculate the importance of features efficiently, i.e. to reduce computational complexity. Comparison of ShapG with other existing XAI methods shows that it provides more accurate explanations for two examined datasets. We also compared other XAI methods developed based on cooperative game theory with ShapG in running time, and the results show that ShapG exhibits obvious advantages in its running time, which further proves efficiency of ShapG. In addition, extensive experiments demonstrate a wide range of applicability of the ShapG method for explaining complex models. We find ShapG an important tool in improving explainability and transparency of AI systems and believe it can be widely used in various fields. (Joint work with Chi Zhao and Jing Liu.)

Dynamic Games and Applications Seminar

Welfare Effects of Prey-Refuge in Fisheries

Nov 28, 2024 11:00 AM — 12:00 PM (Montreal time)

Zoom webinar link

In this paper, I use a tractable predator-prey model with endogenous harvesting to assess the impact of a prey refuge on fishery performance. Using a two-stage game framework, where the prey-refuge consistently protects a portion of the environment from predators, this paper investigates: $(i)$ how fishers modify their behavior in the presence of a prey-refuge and $(ii)$ the conditions under which the prey-refuge enhances the social welfare. The results show that reducing the intensity of species interactions via the prey-refuge diminishes fishing pressure on both prey and predator populations. Interestingly, although full prey protection maximizes the payoff from prey harvesting, it does not necessarily minimize the predator’s fishing payoff. Necessary and sufficient conditions for the existence of positive cooperative surplus are provided. Numerical examples reveal that the overall efficiency of the fishery, influenced by the prey-refuge, is highly contingent on fishers’ willingness to wait for its benefits (i.e., the discount factor). Specifically, when fishers are sufficiently patient, the prey-refuge improves social welfare. Finally, prey refuge implementation can also occur when transfers are not allowed, or when fishers coordinate their fishing strategy. This paper contributes to the fishery management literature by proposing an alternative approach that can promote efficiency through indirect incentives.

Dynamic Games and Applications Seminar

A Differential Game Approach for the Opioid Epidemic

Nov 21, 2024 11:00 AM — 12:00 PM

Zoom webinar link

Under the continuous evolution of the opioid epidemic in the US, the recent increase in addiction as well as overdose deaths due to synthetic opioids, despite increased regulations on prescription practices, warrant a more long-term forward-looking approach incorporating the strategic responses of the other stakeholders. In this work, we model the interactions between the multiple stakeholders, as a three player non-cooperative differential game between the government, health-care providers and the illicit drug providers. The state dynamics for our differential game approach is built upon a recent data-driven epidemiological model on the opioid addiction in Tennessee, extending it to include the effects of the controls or strategies of the three players. In this paper, we compute the open loop Nash Equilibrium using the minimum principle and analyze the equilibrium control strategies of the three players and compare the optimal state trajectory of the system with that of the uncontrolled system.