Values and inductive risk in machine learning modelling: the case of binary classification models

I examine the construction and evaluation of machine learning (ML) binary classification models. These models are increasingly used for societal applications such as classifying patients into two categories according to the presence or absence of a certain disease like cancer and heart disease. I argue that the construction of ML (binary) classification models involves an optimisation process aiming at the minimization of the inductive risk associated with the intended uses of these models. I also argue that the construction of these models is underdetermined by the available data, and that this makes it necessary for ML modellers to make social value judgments in determining the error costs (associated with misclassifications) used in ML optimization. I thus suggest that the assessment of the inductive risk with respect to the social values of the intended users is an integral part of the construction and evaluation of ML classification models. I also discuss the implications of this conclusion for the philosophical debate concerning inductive risk.

Agent-Based Modeling and Simulation of Tourism Market Recovery Strategy after COVID-19 in Yunnan, China

The tourism industry hit severely by COVID-19 faces the challenge of developing effective market recovery strategies. Nonetheless, the existing literature is still limited regarding the dynamic evolution process and management practice. Hence, this study chose several famous spots in the Yunnan Province of China as the focus for a case study and utilized an agent-based simulation method for the decision-making process of tourists’ destination selection and the dynamic recovery process of the destinations under different price and information strategies. The study found that the recovery effects of information strategies are positive, negative, or have no effect in different destinations. In contrast, price strategies can significantly stimulate an increase in the market share of destinations. When price strategy and information strategy are applied simultaneously, the interaction effects are inconsistent in different destinations. The findings contribute to the prediction of the recovery effect of strategies, can reduce trial and error costs, and can improve the scientific understanding of tourism market recovery.

Avoiding costly mistakes in groups: The evolution of error management in collective decision making

Balancing the costs of alternative decisions is a fundamental challenge for decision makers. This is especially critical in social situations, where the choices individuals face are often associated with highly asymmetric error costs---such as pedestrian groups crossing the street, police squads holding a suspect at gunpoint, or animal groups evading predation. While a broad literature has explored how individuals acting alone adapt to asymmetric error costs, little is known about how individuals in groups cope with these costs. Here we investigate adaptive decision strategies of individuals in groups facing asymmetric error costs, modeling scenarios where individuals aim to maximize group-level payoff (‘‘cooperative groups’’) or individual-level payoff (‘‘competitive groups’’). We extended the drift--diffusion model to the social domain in which individuals first gather personal information independently; they can then either wait for additional social information or decide early, thereby potentially influencing others. We combined this social drift--diffusion model with an evolutionary algorithm to derive adaptive behavior. Under asymmetric costs, small cooperative groups evolved response biases to avoid the costly error. Large cooperative groups, however, did not evolve response biases, since the danger of response biases triggering false information cascades increases with group size. We show that individuals in competitive groups face a social dilemma: They evolve higher response biases and wait for more information, thereby undermining group performance. Our results have broad implications for understanding social dynamics in situations with asymmetric costs, such as crowd panics and predator detection.

Sex-specific speed-accuracy tradeoffs shape neural processing of acoustic signals

Speed-accuracy tradeoffs – being fast at the risk of being wrong – are fundamental to many decisions and natural selection is expected to resolve these tradeoffs according to the costs and benefits of behavior. We here test the prediction that females and males should integrate information from courtship signals differently because they experience different payoffs along the speed-accuracy continuum. We fitted a neural model of decision making to behavioral data to determine the parameters of temporal integration of acoustic directional information used by male grasshoppers to locate receptive females. The model revealed that males had a low threshold for initiating a turning response, yet a large integration time constant enabled them to continue to gather information when cues were weak. This contrasts with parameters estimated for females of the same species when evaluating potential mates, in which thresholds for response were much higher and behavior was strongly influenced by unattractive stimuli. Our results reveal differences in neural integration consistent with the sex-specific costs of mate search: Males often face competition and need to be fast, females often pay high error costs and need to be deliberate.Graphical abstract

Through the Anti-Monopoly Lens: What Constitutes ‘Unfairly High Patent Pricing’ in China?

This article proposes that a patentee’s unilateral pricing of proprietary technology should be presumed legal per se under Sec. 55 IPR immunity framework provided by the Anti-Monopoly Law, unless a plaintiff overcomes all three of the following hurdles with actual evidence: (i) the patentee enjoys a dominant market position; (ii) such pricing constitutes de facto refusal to deal with or significant ‘margin squeeze’ for subsequent or follow-on innovators; and (iii) the constructive refusal or ‘margin squeeze’ would likely foreclose dynamic competition. This seemingly high evidentiary burden is justified by three cumulative resources: (i) the very patent mechanism in facilitating innovation, including a solid promise of supra-competitive profit through the right to lawfully exclude competition by imitation, and thus the instigation of a virtuous circle of dynamic competition through pivoting on the critical link of competition by substitution; (ii) the prevalent cautious attitude in sister jurisdictions when dealing with the concept; and (iii) the inevitable limitations of antitrust law, manifested in the administrative and error costs due to lack of proper information and economic analysis methodologies on dynamic efficiency. Through a detailed illustration with six specific scenarios, we see in a quasi-quantitative way that the actual likelihood of unilateral foreclosure on dynamic competition, even in the case of a monopolist patentee, is extremely low despite the existence of a theoretical possibility. Facing this meager likelihood and information deficiency, it would be unwise for a Chinese court to incur enormous costs of searching for a possibility in every case, with the mere guidance of a vague rule-of-reason framework.