What-If XAI Framework (WiXAI): From Counterfactuals towards Causal Understanding

People learn causal relations since childhood using counterfactual reasoning. Counterfactual reasoning uses counterfactual examples which take the form of “what if this has happened differently”. Counterfactual examples are also the basis of counterfactual explanation in explainable artificial intelligence (XAI). However, a framework that relies solely on optimization algorithms to find and present counterfactual samples cannot help users gain a deeper understanding of the system. Without a way to verify their understanding, the users can even be misled by such explanations. Such limitations can be overcome through an interactive and iterative framework that allows the users to explore their desired “what-if” scenarios. The purpose of our research is to develop such a framework. In this paper, we present our “what-if” XAI framework (WiXAI), which visualizes the artificial intelligence (AI) classification model from the perspective of the user’s sample and guides their “what-if” exploration. We also formulated how to use the WiXAI framework to generate counterfactuals and understand the feature-feature and feature-output relations in-depth for a local sample. These relations help move the users toward causal understanding.

Embodied Meaning in Comprehending Abstract Chinese Counterfactuals

Embodied cognition theories propose that language comprehension triggers a sensorimotor system in the brain.However,most previous research has paid much attention to concrete and factual sentences,and little emphasis has been put on the research of abstract and counterfactual sentences.The primary challenges for embodied theories lie in elucidating the meanings of abstract and counterfactual sentences.The most prevalent explanation is that abstract and counterfactual sentences are grounded in the activation of a sensorimotor system,in exactly the same way as concrete and factual ones.The present research employed a dual-task experimental paradigm to investigate whether the embodied meaning is activated in comprehending action-related abstract Chinese counterfactual sentences through the presence or absence of action-sentence compatibility effect(ACE).Participants were instructed to read and listen to the action-related abstract Chinese factual or counterfactual sentences describing an abstract transfer word towards or away from them,and then move their fingers towards or away from them to press the buttons in the same direction as the motion cue of the transfer verb.The action-sentence compatibility effect was observed in both abstract factual and counterfactual sentences,in line with the embodied cognition theories,which indicated that the embodied meanings were activated in both action-related abstract factuals and counterfactuals.

Graph CA: Learning From Graph Counterfactual Augmentation for Knowledge Tracing

With the popularity of online learning in educational settings, knowledge tracing(KT) plays an increasingly significant role. The task of KT is to help students learn more effectively by predicting their next mastery of knowledge based on their historical exercise sequences. Nowadays, many related works have emerged in this field, such as Bayesian knowledge tracing and deep knowledge tracing methods. Despite the progress that has been made in KT, existing techniques still have the following limitations: 1) Previous studies address KT by only exploring the observational sparsity data distribution, and the counterfactual data distribution has been largely ignored. 2) Current works designed for KT only consider either the entity relationships between questions and concepts, or the relations between two concepts, and none of them investigates the relations among students, questions, and concepts, simultaneously, leading to inaccurate student modeling. To address the above limitations,we propose a graph counterfactual augmentation method for knowledge tracing. Concretely, to consider the multiple relationships among different entities, we first uniform students, questions, and concepts in graphs, and then leverage a heterogeneous graph convolutional network to conduct representation learning.To model the counterfactual world, we conduct counterfactual transformations on students’ learning graphs by changing the corresponding treatments and then exploit the counterfactual outcomes in a contrastive learning framework. We conduct extensive experiments on three real-world datasets, and the experimental results demonstrate the superiority of our proposed Graph CA method compared with several state-of-the-art baselines.

Security proof of counterfactual quantum cryptography against general intercept-resend attacks and its vulnerability

Counterfactual quantum cryptography, recently proposed by Noh, is featured with no transmission of signal parti- cles. This exhibits evident security advantages, such as its immunity to the well-known photon-number-splitting attack. In this paper, the theoretical security of counterfactual quantum cryptography protocol against the general intercept- resend attacks is proved by bounding the information of an eavesdropper Eve more tightly than in Yin＇s proposal [Phys. Rev. A 82 042335 （2010）]. It is also shown that practical counterfactual quantum cryptography implementations may be vulnerable when equipped with imperfect apparatuses, by proving that a negative key rate can be achieved when Eve launches a time-shift attack based on imperfect detector efficiency.

Confounding of Three Binary-Variable Counterfactual Model with DAG

Confounding of three binary-variable counterfactual model with directed acyclic graph (DAG) is discussed in this paper. According to the effect between the control variable and the covariate variable, we investigate three causal counterfactual models: the control variable is independent of the covariate variable, the control variable has the effect on the covariate variable and the covariate variable affects the control variable. Using the ancillary information based on conditional independence hypotheses and ignorability, the sufficient conditions to determine whether the covariate variable is an irrelevant factor or whether there is no confounding in each counterfactual model are obtained.

Logical Difficulty from Combining Counterfactuals in the GHZ-Bell Theorems

In eliminating the fair sampling assumption, the Greenberger, Horne, Zeilinger (GHZ) theorem is believed to confirm Bell’s historic conclusion that local hidden variables are inconsistent with the results of quantum mechanics. The GHZ theorem depends on predicting the results of sets of measurements of which only one may be performed. In the present paper, the noncommutative aspects of these unperformed measurements are critically examined. Classical examples and the logic of the GHZ construction are analyzed to demonstrate that combined counterfactual results of noncommuting operations are in general logically inconsistent with performed measurement sequences whose results depend on noncommutation. The Bell theorem is also revisited in the light of this result. It is concluded that negative conclusions regarding local hidden variables do not follow from the GHZ and Bell theorems as historically reasoned.

The Effect of Counterfactual Thinking on the Sunk Cost Effect

Counterfactual thinking is helpful to comprehend the mistakes made previously and to move toward by proposing future actions to facilitate the success of self-regulation.The sunk cost effect rationalizes a strategy,which is the aim of studies on decision-making.However,few of them have discussed the influence of counterfactual thinking to the sunk cost effect.This study assumes that downward counterfactual thinking can regulate the unhappy mood at the moment for relief,which may reduce the sunk cost fallacy;upward counterfactual thinking,on the contrary,emphasizes the improvement of future behaviors,which may increase the sunk cost fallacy.

Chinese Native Speakers' Counterfactuals Revisited

This paper examines whether or not Chinese native speakers (CNSs) have difficulties in understanding English counterfactuals, whether CNSs have counterfactual reasoning problems in their own language, what the causes of these difficulties may be, and the problems in teaching English subjunctives. It also proposes on how to improve CNSs’ English counterfactual comprehension.

Practical Stabilization of Counterfactual Quantum Cryptography

A novel counterfactual quantum key distribution scheme was proposed by T.-G. Noh and a strict security analysis has been given by Z.-Q.Yin, in which two legitimate geographical separated couples may share secret keys even when the key carriers are not traveled in the quantum channel. However, there are still plenty of practical details in this protocol that haven’t been discussed yet, which are of significant importance in physical implementation. In this paper, we will give a practical analysis on such kind of counterfactual quantum cryptography in the aspects of quantum bit error rate (QBER) and stabilization. Furthermore, modified schemes are proposed, which can obtain lower QBER and will be much more robust on stabilization in physical implementation.

Analyses of Political Crisis Impact on Tourism:A Panel Counterfactual Approach with Internet Search Index

Existing research has shown that political crisis events can directly impact the tourism industry.However,the current methods suffer from potential changes of unobserved variables,which poses challenges for a reliable evaluation of the political crisis impacts.This paper proposes a panel counterfactual approach with Internet search index,which can quantitatively capture the change of crisis impacts across time and disentangle the effect of the event of interest from the rest.It also provides a tool to examine potential channels through which the crisis may affect tourist outflows.This research empirically applies the framework to analyze the THAAD event on tourist flows from the Chinese Mainland to South Korea.Findings highlight the strong and negative short-term impact of the political crisis on the tourists' intentions to visit a place.This paper provides essential evidence to help decision-makers improve the management of the tourism crisis.

Quantum counterfactuality with identical particles

Quantum self-interference enables the counterfactual transmission of information,whereby the transmitted bits involve no particles traveling through the channel.In this work,we show how counterfactuality can be realized even when the self-interference is replaced by interference between identical particles.Interestingly,the facet of indistinguishability called forth here is associated with first-order coherence,and is different from the usual notion of indistinguishability associated with the(anti-)commutation relations of mode operators.From an experimental perspective,the simplest implementation of the proposed idea can be realized by slight modifications to existing protocols for differential-phase-shift quantum key distribution or interaction-free measurement.

Counterfactual Reasoning over Community Detection:A Case Study of the Public Science Day Community

With the rapid rise of new media platforms such as Weibo and Tiktok,communities with science communication characteristics have progressively grown on social networks.These communities pursue essential objectives such as increased visibility and influence.For the success of the public understanding of science in China,case studies of science communication communities on social media are becoming increasingly valuable as a point of reference.The authenticity of user influence plays an important role in the analysis of the final outcome during the process of community detection.By integrating counterfactual reasoning theory into a community detection algorithm,we present a novel paradigm for eliminating influence bias in online communities.We consider the community of Public Science Day of the Chinese Academy of Sciences as a case study to demonstrate the validity of the proposed paradigm.In addition,we examine data on science communication activities,analyze the key elements of activity communication,and provide references for not only augmenting the communication impact of similar types of popular science activities but also advancing science communication in China.Our main finding is that the propagation channel for the science communication experiment exhibits multi-point scattered propagation and lacks a continuous chain in the process of propagation.

Influence of ecological function protection zone on the water conservation in Gansu-Qinghai Contiguous Region of the upper Yellow River

The implementation of Ecological Function Protection Zone(EFPZ)policy is significant for the ecological restoration and conservation of soil and water in the territory space.This manuscript analyzed and quantified the impact of EFPZ on the regional water conservation function,based on land use data from 2005,2008,2010,2015 and 2020,by conducting a counterfactual simulation along with the GeoSOS-FLUS model and the InVEST model.The results demonstrate that the delineation of EFPZ can significantly influence the water conservation.(1)From 2010 to 2020,as the EFPZ was implemented,the water conservation in the study area was increasing year by year,with a growth rate of 0.03×10^(8) m^(3)∙a^(-1).On the other hand,the simulated water conservation capacity without the implementation of EFPZ decreased year by year,with a decrease rate of 0.01×10^(8) m^(3)∙a^(-1).(2)The EFPZ accounts for only 23%of the total area,but the contribution rate of water conservation reaches 80%.The actual values of water conservation and average water yield per unit pixel in the EFPZ show an increasing trend both internally and externally,while the counterfactual simulation values exhibit a decreasing trend.(3)The water conservation is much higher within the EFPZ than without EFPZ.The implementation of EFPZ has a significant effect on the improvement of the water conservation capacity in Maqu EFPZ and Yellow River Source EFPZ.The protection effectiveness should be enhanced in Qilian Mountain EFPZ and afforestation activities need to be carefully considered in Loess Plateau EFPZ.

Counterfactuals and Context-Sensitivity

It is commonly agreed that when evaluating the validity of an argument involving context-sensitive expressions, the context should be held fixed. In their 2008 essay 'Counterfactuals and Context,' Brogaard and Salerno argue further that context should be held fixed when evaluating an argument involving counterfactuals for validity, since, as many will agree, counterfactuals are context-sensitive. In the present paper, it will however be argued that Brogaard and Salerno fail to distinguish between two different roles that context plays in determining the meaning of a given counterfactual. If they were fully aware of the distinction between these two roles played by context, they might propose a contextualist approach to counterfactuals, as has been developed by Ichikawa in his 2011 paper 'Quantifiers, Knowledge, and Counterfactuals.'

Causal Inference

Causal inference is a powerful modeling tool for explanatory analysis,which might enable current machine learning to become explainable.How to marry causal inference with machine learning to develop explainable artificial intelligence(XAI)algorithms is one of key steps toward to the artificial intelligence 2.0.With the aim of bringing knowledge of causal inference to scholars of machine learning and artificial intelligence,we invited researchers working on causal inference to write this survey from different aspects of causal inference.This survey includes the following sections:“Estimating average treatment effect:A brief review and beyond”from Dr.Kun Kuang,“Attribution problems in counterfactual inference”from Prof.Lian Li,“The Yule–Simpson paradox and the surrogate paradox”from Prof.Zhi Geng,“Causal potential theory”from Prof.Lei Xu,“Discovering causal information from observational data”from Prof.Kun Zhang,“Formal argumentation in causal reasoning and explanation”from Profs.Beishui Liao and Huaxin Huang,“Causal inference with complex experiments”from Prof.Peng Ding,“Instrumental variables and negative controls for observational studies”from Prof.Wang Miao,and“Causal inference with interference”from Dr.Zhichao Jiang.

Cooperative and Competitive Multi-Agent Systems:From Optimization to Games

Multi-agent systems can solve scientific issues related to complex systems that are difficult or impossible for a single agent to solve through mutual collaboration and cooperation optimization.In a multi-agent system,agents with a certain degree of autonomy generate complex interactions due to the correlation and coordination,which is manifested as cooperative/competitive behavior.This survey focuses on multi-agent cooperative optimization and cooperative/non-cooperative games.Starting from cooperative optimization,the studies on distributed optimization and federated optimization are summarized.The survey mainly focuses on distributed online optimization and its application in privacy protection,and overviews federated optimization from the perspective of privacy protection me-chanisms.Then,cooperative games and non-cooperative games are introduced to expand the cooperative optimization problems from two aspects of minimizing global costs and minimizing individual costs,respectively.Multi-agent cooperative and non-cooperative behaviors are modeled by games from both static and dynamic aspects,according to whether each player can make decisions based on the information of other players.Finally,future directions for cooperative optimization,cooperative/non-cooperative games,and their applications are discussed.

Assessing the spillover effects of U.S.monetary policy normalization on Nigeria sovereign bond yield

This study examines the spillover effects of U.S.monetary policy normalization on Nigeria 10-Year Treasury bond yield between 2011 and 2017,using the vector error correction model approach.Our results reveal that domestic factors,such as exchange rate and inflation,rather than the U.S.10-Year sovereign bond yield,are the key drivers of Nigeria 10-Year bond yield.Additionally,the spillover effect from the U.S.monetary policy was amplified by oil price shocks and changes in Nigeria’s monetary policy rates.Our counterfactual analysis confirms the findings.

Modelling HIV/AIDS Cases in Zambia: A Comparative Study of the Impact of Mandatory HIV Testing

In this study, a time series modeling approach is used to determine an ARIMA model and advance counterfactual forecasting at a point of policy intervention. We consider monthly data of HIV/AIDS cases from the Ministry of Health (Copperbelt province) of Zambia, for the period 2010 to 2019 and have a total of 120 observations. Results indicate that ARIMA (1, 0, 0) is an adequate model which best fits the HIV/AIDS time series data and is, therefore, suitable for forecasting cases. The model predicts a reduction from an average of 3500 to 3177 representing 14.29% in HIV/AIDS cases from 2017 (year of policy activation) to 2019, but the actual recorded cases dropped from 3500 to 1514 accounting for 57.4% in the same time frame.

A Study on High Incomes in China's Monopoly Industries

High incomes in state monopoly industries have drawn extensive public concerns in China. The management of state-owned enterprises(SOEs) is motivated and empowered to set payment schemes in their own favor, resulting in the greater extent to which their senior executives are overpaid as compared with average workers. Hence, the level of unjustified high incomes is heterogeneous among employees of different income levels in state monopoly industries. Using Machado-Mata counterfactual decomposition method based on the regression of multiple quantiles, this paper measures the proportions of reasonable and unreasonable parts of pay gap between state monopoly industries and competitive industries. Our study found that the workforce of state monopoly industries are overpaid across various wage levels, while the extent to which they are overpaid increases with the rise of wage level. The implication is that compared with average workers, executives in state monopoly industries are overpaid to an even greater extent. This requires that the government focus on curbing the high executive income in addition to putting a lid on the gross payroll of SOEs in state monopoly industries. The fundamental strategy to resolving the excessive high income in state monopoly industries is to bring different types of SOEs under different corporate management models and income systems.

“本来”反事实句与情态共现问题研究

文章从对"本来"的反事实语义考察出发,探讨"本来"反事实句与情态共现的问题。我们发现,如果以事物的本来性状作为参照物,那么"本来"句具有三种可能语义关系:正向关系、零向关系和反向关系。正向关系不能产生反事实语义。零向关系中,"本来"必须借助情态动词的帮助形成"本来+就+情态动词"结构才可能产生反事实语义。其中情态类型最有可能为道义情态。反向关系中,根据情态动词是否参与反事实解读,"本来"反事实语义可能有两种不同类型。并且不同情态类型在"本来"反事实句中的表现也不均衡:道义情态的使用最为受限,动力情态的使用用例较少,认识情态的使用最为自由。