While deep learning methods have achieved strong performance in time series prediction, their black-box nature and inability to explicitly model underlying stochastic processes often limit their robustness handling non-stationary data, especially in the presence of abrupt changes. In this work, we introduce Neural MJD, a neural network based non-stationary Merton jump diffusion (MJD) model. Our model explicitly formulates forecasting as a stochastic differential equation (SDE) simulation problem, combining a time-inhomogeneous Itô diffusion to capture non-stationary stochastic dynamics with a time-inhomogeneous compound Poisson process to model abrupt jumps. To enable tractable learning, we introduce a likelihood truncation mechanism that caps the number of jumps within small time intervals and provide a theoretical error bound for this approximation. Additionally, we propose an Euler-Maruyama with restart solver, which achieves a provably lower error bound in estimating expected states and reduced variance compared to the standard solver. Experiments on both synthetic and real-world datasets demonstrate that Neural MJD consistently outperforms state-of-the-art deep learning and statistical learning methods. Our code is available at https://github.com/DSL-Lab/neural-MJD.

Large language models (LLMs) now stand in for human decision makers in many settings, but it is unclear whether their choices arise from structured internal representations or surface-level pattern matching—and whether the economic preferences they express are correlated as in humans. We address these questions by analyzing three canonical preference domains - risk, time, and social - across several current LLMs.

In network games, individuals interact strategically within network environments to maximize their utilities. However, obtaining network structures is challenging. In this work, we propose an unsupervised learning model, called data-dependent gated-prior graph variational autoencoder (GPGVAE), that infers the underlying latent interaction type (strategic complement vs. substitute) among individuals and the latent network structure based on their observed actions. Specially, we propose a spectral graph neural network (GNN) based encoder to predict the interaction type and a data-dependent gated prior that models network structures conditioned on the interaction type. We further propose a Transformer based mixture of Bernoulli encoder of network structures and a GNN based decoder of game actions. We systematically study the Monte Carlo gradient estimation methods and effectively train our model in a stage-wise fashion. Extensive experiments across various synthetic and real-world network games demonstrate that our model achieves state-of-the-art performances in inferring network structures and well captures interaction types.

We examine the economic decision-making and behavioral biases of Large Language Models (LLMs) across English, Chinese, Spanish, and French, with a specific emphasis on mental accounting. We develop a probabilistic graphical model to dissect the influences of reasoning factors on LLMs' decision-making processes.

Decision-making on networks can be explained by both homophily and social influences. While homophily drives the formation of communities with similar characteristics, social influences occur both within and between communities. Social influences can be reasoned through role theory, which indicates that the influences among individuals depending on their roles and the behavior of interest. To operationalize these social science theories, we empirically identify the homophilous communities and use the community structures to capture such “roles”, affecting particular decision-making processes. We propose a generative model named the Stochastic Block Influence Model and jointly analyze both network formation and behavioral influences within and between different empirically-identified communities. To evaluate the performance and demonstrate the interpretability of our method, we study the adoption decisions for a microfinance product in Indian villages. We show that although individuals tend to form links within communities, there are strongly positive and negative social influences between communities, supporting the weak ties theory. Moreover, communities with shared characteristics are associated with positive influences. In contrast, communities that do not overlap are associated with negative influences. Our framework facilitates the quantification of the influences underlying decision communities and is thus a helpful tool for driving information diffusion, viral marketing, and technology adoption.

Strategic interactions between a group of individuals or organisations can be modelled as games played on networks, where a player’s payoff depends not only on their actions but also on those of their neighbours. Inferring the network structure from observed game outcomes (equilibrium actions) is an important problem with numerous potential applications in economics and social sciences. Existing methods mostly require the knowledge of the utility function associated with the game, which is often unrealistic to obtain in real-world scenarios. We adopt a transformer-like architecture which correctly accounts for the symmetries of the problem and learns a mapping from the equilibrium actions to the network structure of the game without explicit knowledge of the utility function. We test our method on three different types of network games using both synthetic and real-world data, and demonstrate its effectiveness in network structure inference and superior performance over existing methods.

Automatic epilepsy detection is of great significance for the diagnosis and treatment of patients. Most detection methods are based on patient-specific models and have achieved good results. However, in practice, new patients do not have their own previous EEG data and therefore cannot be initially diagnosed. If the EEG data of other patients can be used to achieve cross-patient detection, and cross-patient and patient-specific experiments can be combined at the same time, this method will be more widely used. In this work, an EEG classification model based on a self-organizing fuzzy logic (SOF) classifier is proposed for both cross-patient and patient-specific seizure detection. After preprocessing, the features of the original EEG signal are extracted and sent to the SOF classifier. This classification model is free from predefined parameters or a prior assumption regarding the EEG data generation model and only stores the key meta-parameters in memory. Therefore, it is very suitable for large-scale EEG signals in cross-patient detection. Selecting different granularity and classification distance in two different experiments after post-processing will achieve the best results. Experiments were conducted using a long-term continuous scalp EEG database and the [Formula: see text]-mean of cross-patient and patient-specific detection reached 83.35% and 92.04%, respectively. A comparison with other methods shows that there is greater performance and generalizability with this method.

Abstract Scientific novelty drives the efforts to invent new vaccines and solutions during the pandemic. First‐time collaboration and international collaboration are two pivotal channels to expand teams' search activities for a broader scope of resources required to address the global challenge, which might facilitate the generation of novel ideas. Our analysis of 98,981 coronavirus papers suggests that scientific novelty measured by the BioBERT model that is pretrained on 29 million PubMed articles, and first‐time collaboration increased after the outbreak of COVID‐19, and international collaboration witnessed a sudden decrease. During COVID‐19, papers with more first‐time collaboration were found to be more novel and international collaboration did not hamper novelty as it had done in the normal periods. The findings suggest the necessity of reaching out for distant resources and the importance of maintaining a collaborative scientific community beyond nationalism during a pandemic.

Location-Based Social Networks have been widely studied in recent years; new approaches constantly developed to solve individuals’ trajectory prediction tasks. However, most of these methods require sufficient data to learn individual features, which is not always satisfied in real situations, especially for online data. The digital data on human behavior typically follows a power-law distribution, indicating that only a few people have rich activities recorded while most people’s behavioral data are limited. In order to overcome this hurdle, our work constructs the user behavior proximity network (UBPN) and proposes a new walking strategy based on this network that extracts the hidden information from the social contacts to substitute the unobserved behavioral information of an individual. Specifically, our proposed walking strategy has two walking paths, accounting for the temporal and social information on the ego users’ and their alters’ mobility activities. This walking strategy is model-agnostic and can be integrated with many existing walk-based deep learning methods. Our work applies the methods on two real-world datasets with rich spatiotemporal information and shows that the performances of the existing prediction methods improve significantly by integrating the proposed walking strategy.

Not only did COVID-19 give rise to a global pandemic, but also it resulted in an infodemic comprising misinformation, rumor, and propaganda. The consequences of this infodemic can erode public trust, impede the containment of the virus, and outlive the pandemic itself. The evolving and fragmented media landscape, particularly the extensive use of social media, is a crucial driver of the spread of misinformation. Focusing on the Chinese social media Weibo, we collected four million tweets, from December 9, 2019, to April 4, 2020, examining misinformation identified by the fact-checking platform Tencent-a leading Chinese tech giant. Our results show that the evolution of misinformation follows an issue-attention cycle pertaining to topics such as city lockdown, cures and preventive measures, school reopening, and foreign countries. Sensational and emotionally reassuring misinformation characterizes the whole issue-attention cycle, with misinformation on cures and prevention flooding social media. We also study the evolution of sentiment and observe that positive sentiment dominated over the course of Covid, which may be due to the unique characteristic of "positive energy" on Chinese social media. Lastly, we study the media landscape during Covid via a case study on a controversial unproven cure known as Shuanghuanglian, which testifies to the importance of scientific communication in a plague. Our findings shed light on the distinct characteristics of misinformation and its cultural, social, and political implications, during the COVID-19 pandemic. The study also offers insights into combating misinformation in China and across the world at large.

Individuals, or organizations, cooperate with or compete against one another in a wide range of practical situations. Such strategic interactions are often modeled as games played on networks, where an individual’s payoff depends not only on her action but also on that of her neighbors. The current literature has largely focused on analyzing the characteristics of network games in the scenario where the structure of the network, which is represented by a graph, is known beforehand. It is often the case, however, that the actions of the players are readily observable while the underlying interaction network remains hidden. In this paper, we propose two novel frameworks for learning, from the observations on individual actions, network games with linear-quadratic payoffs, and in particular, the structure of the interaction network. Our frameworks are based on the Nash equilibrium of such games and involve solving a joint optimization problem for the graph structure and the individual marginal benefits. Both synthetic and real-world experiments demonstrate the effectiveness of the proposed frameworks, which have theoretical as well as practical implications for understanding strategic interactions in a network environment.

The automatic identification of epileptic electroencephalogram (EEG) signals can give assistance to doctors in diagnosis of epilepsy, and provide the higher security and quality of life for people with epilepsy. Feature extraction of EEG signals determines the performance of the whole recognition system. In this paper, a novel method using the local binary pattern (LBP) based on the wavelet transform (WT) is proposed to characterize the behavior of EEG activities. First, the WT is employed for time-frequency decomposition of EEG signals. After that, the "uniform" LBP operator is carried out on the wavelet-based time-frequency representation. And the generated histogram is regarded as EEG feature vector for the quantification of the textural information of its wavelet coefficients. The LBP features coupled with the support vector machine (SVM) classifier can yield the satisfactory recognition accuracies of 98.88% for interictal and ictal EEG classification and 98.92% for normal, interictal and ictal EEG classification on the publicly available EEG dataset. Moreover, the numerical results on another large size EEG dataset demonstrate that the proposed method can also effectively detect seizure events from multi-channel raw EEG data. Compared with the standard LBP, the "uniform" LBP can obtain the much shorter histogram which greatly reduces the computational burden of classification and enables it to detect ictal EEG signals in real time.

Audio is an important part of multimedia, and it has many useful applications in real life. Audio event classification is a key technology in audio management and application. Supervised audio event classification requires labeling large amounts of samples, while manual labeling is a very time-consuming work. In this paper we propose SVMCKNND, an active learning method for SVM classifier, to deal with the labeling problem in audio event classification. For SVMCKNND, in each iteration, first, a low-confidence region is delimited; then based on KNN, the samples that are more likely to be on the true class boundary are taken as the informative ones; finally, redundancy that exists in the informative samples is reduced to further decrease manual labeling workload. Experimental results show that SVMCKNNDperforms better than another two SVM active learning algorithms, especially in classifying small-sample audio events.

A new feature extraction method for face recognition based on principal component analysis (PCA) and fractional-step linear discriminant analysis (F-LDA) is given in this paper. In order to reduce the computation complexity, PCA is first used to reduce the dimension. In addition, before using F-LDA, we transform the pooled within-class scatter matrix into an identity matrix. The proposed method is tested on AR and UMIST face databases. Experiment results show that our method gains higher classification accuracy than other existing methods used in the experiment.