Estimating causal effects from observational network data is a significant but challenging problem. Existing works in causal inference for observational network data lack an analysis of the generalization bound, which can theoretically provide support for alleviating the complex confounding bias and practically guide the design of learning objectives in a principled manner. To fill this gap, we derive a generalization bound for causal effect estimation in network scenarios by exploiting 1) the reweighting schema based on joint propensity score and 2) the representation learning schema based on Integral Probability Metric (IPM). We provide two perspectives on the generalization bound in terms of reweighting and representation learning, respectively. Motivated by the analysis of the bound, we propose a weighting regression method based on the joint propensity score augmented with representation learning. Extensive experimental studies on two real-world networks with semi-synthetic data demonstrate the effectiveness of our algorithm.

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 task using an algebraic characterization of directed acyclic graphs (DAGs) and the least-square loss function. Though the least-square loss function is well justified under the standard Gaussian noise assumption, it is limited if the assumption does not hold. In this work, we theoretically show that the violation of the Gaussian noise assumption will hinder the causal direction identification, making the causal orientation fully determined by the causal strength as well as the variances of noises in the linear case and by the strong non-Gaussian noises in the nonlinear case. Consequently, we propose a more general entropy-based loss that is theoretically consistent with the likelihood score under any noise distribution. We run extensive empirical evaluations on both synthetic data and real-world data to validate the effectiveness of the proposed method and show that our method achieves the best in structure Hamming distance, false discovery rate (FDR), and true-positive rate (TPR) matrices.

Estimating long-term causal effects by combining long-term observational and short-term experimental data is a crucial but challenging problem in many real-world scenarios. In existing methods, several ideal assumptions, e.g. latent unconfoundedness assumption or additive equi-confounding bias assumption, are proposed to address the latent confounder problem raised by the observational data. However, in real-world applications, these assumptions are typically violated which limits their practical effectiveness. In this paper, we tackle the problem of estimating the long-term individual causal effects without the aforementioned assumptions. Specifically, we propose to utilize the natural heterogeneity of data, such as data from multiple sources, to identify latent confounders, thereby significantly avoiding reliance on idealized assumptions. Practically, we devise a latent representation learning-based estimator of long-term causal effects. Theoretically, we establish the identifiability of latent confounders, with which we further achieve long-term effect identification. Extensive experimental studies, conducted on multiple synthetic and semi-synthetic datasets, demonstrate the effectiveness of our proposed method.

Foundation models enable the reuse of pretrained knowledge across time series classification (TSC) tasks, but transfer to a new task still requires target information to establish the task specific decision rule. Existing surveys mainly organize time series foundation models by architecture, pretraining objective, training data, or source modality, while the target classification procedure receives less attention. We instead organize foundation models for TSC into three transfer mechanisms: representation transfer, fine-tuning and adaptation, and in-context learning. Alongside the survey, we develop and release the TSC-FM Benchmark, a unified evaluation infrastructure covering 198 TSC datasets and multiple target settings. We use the benchmark to study how downstream classifiers, input settings, and task characteristics affect the assessment of pretrained models. The results show that the same pretrained representation can receive substantially different assessments under different downstream classifiers, that the best reported low shot model differs between univariate and multivariate settings, and that method groups lose accuracy at different rates as the number of classes increases. These findings suggest that transfer capability depends on the match between pretrained knowledge, target task structure, and the target procedure, rather than on the pretrained backbone alone. Future research should both expand the effective range of transfer and reliably identify when a target task falls outside that range.