Domain adaptation is an important but challenging task. Most of the existing domain methods struggle to extract domain-invariant representation on feature space with entangling information and semantic information. Different from previous efforts entangled space, we aim invariant in latent disentangled (DSR) data. In DSR, assume data generation process controlled by two independent sets variables, i.e., variables variables. Under above assumption, employ a variational auto-encoder...

Learning causal structure among event types on multitype sequences is an important but challenging task. Existing methods, such as the Multivariate Hawkes processes, mostly assumed that each sequence independent and identically distributed. However, in many real-world applications, it commonplace to encounter a topological network behind excited or inhibited not only by its history also neighbors. Consequently, failure describing dependency leads error detection of structure. By considering...

Causal discovery without intervention is well recognized as a challenging yet powerful data analysis tool, boosting the development of other scientific areas, such biology, astronomy, and social science. The major technical difficulty behind observation-based causal to effectively efficiently identify causes effects from correlated variables given existence significant noises. Previous studies mostly employ two very different methodologies under Bayesian network framework, namely global...

Deep neural networks (DNNs) have been demonstrated to be vulnerable well-crafted adversarial examples, which are generated through either well-conceived L_p-norm restricted or unrestricted attacks. Nevertheless, the majority of those approaches assume that adversaries can modify any features as they wish, and neglect causal generating process data, is unreasonable unpractical. For instance, a modification in income would inevitably impact like debt-to-income ratio within banking system. By...

Identification of causal direction between a causal-effect pair from observed data has recently attracted much attention. Various methods based on functional models have been proposed to solve this problem, by assuming the process satisfies some (structural) constraints and showing that reverse violates such constraints. The nonlinear additive noise model demonstrated be effective for purpose, but class is not transitive--even if each direct relation follows model, indirect influences, which...

When assessing urban eco-environmental sensitive areas (Urban Eco-ESAs) by multi-criteria evaluation method, the widely used weighted linear combination method may inevitably lead to some factors of high value being neutralized other low value, resulting in neglect eco-sensitive as a consequence, while on hand, Boolean OR which give result long any factor has sensitivity and thus ignore mutual compensation mechanism among ecological factors, can excessively wide ranges areas. To overcome...

Missing data are an unavoidable complication frequently encountered in many causal discovery tasks. When a missing process depends on the values themselves (known as self-masking missingness), recovery of joint distribution becomes unattainable, and detecting presence such missingness remains perplexing challenge. Consequently, due to inability reconstruct original discern underlying mechanism, simply applying existing methods would lead wrong conclusions. In this work, we found that recent...

Conditional independence (CI) testing is an important problem, especially in causal discovery. Most methods assume that all variables are fully observable and then test the CI among observed data. Such assumption often untenable beyond applications dealing with, e.g., psychological analysis about mental health status medical diagnosing (researchers need to consider existence of latent these scenarios); typically adopted schemes mainly suffer from robust or efficient issues. Accordingly, this...

Learning causal structure among event types from discrete-time sequences is a particularly important but challenging task. Existing methods, such as the multivariate Hawkes processes based mostly boil down to learning so-called Granger causality which assumes that cause happens strictly prior its effect event. Such an assumption often untenable beyond applications, especially when dealing with in low-resolution; and typical discrete mainly suffer identifiability issues raised by...

The recommendation system, relying on historical observational data to model the complex relationships among users and items, has achieved great success in real-world applications. Selection bias is one of most important issues existing data-based approaches, which actually caused by multiple types unobserved exposure strategies (e.g., promotions holiday effects). Though various methods have been proposed address this problem, they are mainly implicit debiasing techniques but not explicitly...

Identification of causal direction between a causal-effect pair from observed data has recently attracted much attention. Various methods based on functional models have been proposed to solve this problem, by assuming the process satisfies some (structural) constraints and showing that reverse violates such constraints. The nonlinear additive noise model demonstrated be effective for purpose, but class is not transitive--even if each direct relation follows model, indirect influences, which...

Causal discovery from observational data is an important but challenging task in many scientific fields. A recent line of work formulates the structure learning problem as a continuous constrained optimization using algebraic characterization directed acyclic graphs (DAGs) and least-square loss function. Though function well justified under standard Gaussian noise assumption, it limited if assumption does not hold. In this work, we theoretically show that violation will hinder causal...

Identification of causal direction between a causal-effect pair from observed data has recently attracted much attention. Various methods based on functional models have been proposed to solve this problem, by assuming the process satisfies some (structural) constraints and showing that reverse violates such constraints. The nonlinear additive noise model demonstrated be effective for purpose, but class does not allow any confounding or intermediate variables cause pair–even if each direct...

The explainability of Graph Neural Networks (GNNs) is critical to various GNN applications, yet it remains a significant challenge. A convincing explanation should be both necessary and sufficient simultaneously. However, existing explaining approaches focus on only one the two aspects, necessity or sufficiency, heuristic trade-off between two. Theoretically, Probability Necessity Sufficiency (PNS) holds potential identify most since can mathematically quantify sufficiency an explanation....

Learning Granger causality from event sequences is a challenging but essential task across various applications. Most existing methods rely on the assumption that are independent and identically distributed (i.i.d.). However, this i.i.d. often violated due to inherent dependencies among sequences. Fortunately, in practice, we find these can be modeled by topological network, suggesting potential solution non-i.i.d. problem introducing prior network into causal discovery. This observation...

Count data naturally arise in many fields, such as finance, neuroscience, and epidemiology, discovering causal structure among count is a crucial task various scientific industrial scenarios. One of the most common characteristics inherent branching described by binomial thinning operator an independent Poisson distribution that captures both noise. For instance, population scenario, mortality immigration contribute to count, where survival follows Bernoulli distribution, distribution....

Count data naturally arise in many fields, such as finance, neuroscience, and epidemiology, discovering causal structure among count is a crucial task various scientific industrial scenarios. One of the most common characteristics inherent branching described by binomial thinning operator an independent Poisson distribution that captures both noise. For instance, population scenario, mortality immigration contribute to count, where survival follows Bernoulli distribution, distribution....

Texture represents the surface quality of fabrics, which is one key factors for textiles design. Fabric texture can be perceived via sensory perceptions, like vision and touching, causing different psychological feelings emotions. This paper aimed to model correlation between fabric textures evoked emotions through perceptions. Firstly, 20 subjects were required make evaluation on 10 samples, rating induced visual visual-tactile Then, differences in analyzed. The results showed that only by...

Unobserved discrete data are ubiquitous in many scientific disciplines, and how to learn the causal structure of these latent variables is crucial for uncovering patterns. Most studies focus on linear variable model or impose strict constraints structures, which fail address cases involving non-linear relationships complex structures. To achieve this, we explore a tensor rank condition contingency tables an observed set $\mathbf{X}_p$, showing that determined by minimum support specific...

ADVERTISEMENT RETURN TO ISSUEPREVAddition/CorrectionNEXTORIGINAL ARTICLEThis notice is a correctionAddition to "Bioinspired l-Proline Oligomers for the Cryopreservation of Oocytes via Controlling Ice Growth"Qingyuan QinQingyuan QinMore by Qingyuan Qin, Lishan ZhaoLishan ZhaoMore Zhao, Zhang LiuZhang LiuMore Liu, Tao LiuTao Jiangxue QuJiangxue QuMore Qu, Xiaowei ZhangXiaowei ZhangMore Zhang, Rong LiRong LiMore Li, Liying YanLiying YanMore Yan, Jie Yan*Jie Shenglin Jin*Shenglin JinMore Jin,...

We study the problem of learning hierarchical causal structure among latent variables from measured variables. While some existing methods are able to recover structure, they mostly suffer restricted assumptions, including tree-structured graph constraint, no ``triangle" and non-Gaussian assumptions. In this paper, we relax these restrictions above consider a more general challenging scenario where beyond graph, arbitrary noise distribution allowed. investigate identifiability show that by...

Deep neural networks (DNNs) have been demonstrated to be vulnerable well-crafted \emph{adversarial examples}, which are generated through either well-conceived $\mathcal{L}_p$-norm restricted or unrestricted attacks. Nevertheless, the majority of those approaches assume that adversaries can modify any features as they wish, and neglect causal generating process data, is unreasonable unpractical. For instance, a modification in income would inevitably impact like debt-to-income ratio within...