Yuanchen Bei, Tianxin Wei, Xuying Ning, Yanjun Zhao, Zhining Liu, Xiao Lin, Yada Zhu, Hendrik Hamann, Jingrui He, Hanghang Tong. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Climate science studies the structure and dynamics of Earth's climate system and seeks to understand how climate changes over time, where the data is usually stored in the format of time series, recording the climate features, geolocation, time attributes, etc. Recently, much research attention has been paid to the climate benchmarks. In addition to the most common task of weather forecasting, several pioneering benchmark works are proposed for extending the modality, such as domain-specific applications like tropical cyclone intensity prediction and flash flood damage estimation, or climate statement and confidence level in the format of natural language. To further motivate the artificial intelligence development for climate science, in this paper, we first contribute a multi-modal climate benchmark, i.e., ClimateBench-M, which aligns (1) the time series climate data from ERA5, (2) extreme weather events data from NOAA, and (3) satellite image data from NASA HLS based on a unified spatial-temporal granularity. Second, under each data modality, we also propose a simple but strong generative method that could produce competitive performance in weather forecasting, thunderstorm alerts, and crop segmentation tasks in the proposed ClimateBench-M. The data and code of ClimateBench-M are publicly available at https://github.com/iDEA-iSAIL-Lab-UIUC/ClimateBench-M.

While large language models (LLMs) have made significant progress in processing and reasoning over knowledge graphs, current methods suffer from a high non-retrieval rate. This limitation reduces the accuracy of answering questions based on these graphs. Our analysis reveals that the combination of greedy search and forward reasoning is a major contributor to this issue. To overcome these challenges, we introduce the concept of super-relations, which enables both forward and backward reasoning by summarizing and connecting various relational paths within the graph. This holistic approach not only expands the search space, but also significantly improves retrieval efficiency. In this paper, we propose the ReKnoS framework, which aims to Reason over Knowledge Graphs with Super-Relations. Our framework’s key advantages include the inclusion of multiple relation paths through super-relations, enhanced forward and backward reasoning capabilities, and increased efficiency in querying LLMs. These enhancements collectively lead to a substantial improvement in the successful retrieval rate and overall reasoning performance. We conduct extensive experiments on a variety of datasets to evaluate ReKnoS, and the results demonstrate the superior performance of ReKnoS over existing state-of-the-art baselines, with an average accuracy gain of 2.92% across nine real-world datasets.

Time-series forecasting plays a critical role in many real-world applications. Although increasingly powerful models have been developed and achieved superior results on benchmark datasets, through a fine-grained sample-level inspection, we find that (i) no single model consistently outperforms others across different test samples, but instead (ii) each model excels in specific cases. These findings prompt us to explore how to adaptively leverage the distinct strengths of various forecasting models for different samples. We introduce TimeFuse, a framework for collective time-series forecasting with sample-level adaptive fusion of heterogeneous models. TimeFuse utilizes meta-features to characterize input time series and trains a learnable fusor to predict optimal model fusion weights for any given input. The fusor can leverage samples from diverse datasets for joint training, allowing it to adapt to a wide variety of temporal patterns and thus generalize to new inputs, even from unseen datasets. Extensive experiments demonstrate the effectiveness of TimeFuse in various long-/short-term forecasting tasks, achieving near-universal improvement over the state-of-the-art individual models. Code is available at https://github.com/ZhiningLiu1998/TimeFuse.

Graph mining has become crucial in fields such as social science, finance, and cybersecurity. Many large-scale real-world networks exhibit both heterogeneity, where multiple node and edge types exist in the graph, and heterophily, where connected nodes may have dissimilar labels and attributes. However, existing benchmarks primarily focus on either heterophilic homogeneous graphs or homophilic heterogeneous graphs, leaving a significant gap in understanding how models perform on graphs with both heterogeneity and heterophily. To bridge this gap, we introduce H2GB, a large-scale node-classification graph benchmark that brings together the complexities of both the heterophily and heterogeneity properties of real-world graphs. H2GB encompasses 9 real-world datasets spanning 5 diverse domains, 28 baseline models, and a unified benchmarking library with a standardized data loader, evaluator, unified modeling framework, and an extensible framework for reproducibility. We establish a standardized workflow supporting both model selection and development, enabling researchers to easily benchmark graph learning methods. Extensive experiments across 28 baselines reveal that current methods struggle with heterophilic and heterogeneous graphs, underscoring the need for improved approaches. Finally, we present a new variant of the model, H2G-former, developed following our standardized workflow, that excels at this challenging benchmark. Both the benchmark and the framework are publicly available at Github and PyPI, with documentation hosted at https://junhongmit.github.io/H2GB.

Class-imbalanced learning (CIL) on tabular data is important in many real-world applications where the minority class holds the critical but rare outcomes. In this paper, we present CLIMB, a comprehensive benchmark for class-imbalanced learning on tabular data. CLIMB includes 73 real-world datasets across diverse domains and imbalance levels, along with unified implementations of 29 representative CIL algorithms. Built on a high-quality open-source Python package with unified API designs, detailed documentation, and rigorous code quality controls, CLIMB supports easy implementation and comparison between different CIL algorithms. Through extensive experiments, we provide practical insights on method accuracy and efficiency, highlighting the limitations of naive rebalancing, the effectiveness of ensembles, and the importance of data quality. Our code, documentation, and examples are available at https://github.com/ZhiningLiu1998/imbalanced-ensemble.

Graph heterophily, where connected nodes have different labels, has attracted significant interest recently. Most existing works adopt a simplified approach - using low-pass filters for homophilic graphs and high-pass filters for heterophilic graphs. However, we discover that the relationship between graph heterophily and spectral filters is more complex - the optimal filter response varies across frequency components and does not follow a strict monotonic correlation with heterophily degree. This finding challenges conventional fixed filter designs and suggests the need for adaptive filtering to preserve expressiveness in graph embeddings. Formally, natural questions arise: Given a heterophilic graph $\mathcal{G}$ , how and to what extent will the varying heterophily degree of $\mathcal{G}$ affect the performance of GNNs? How can we design adaptive filters to fit those varying heterophilic connections? Our theoretical analysis reveals that the average frequency response of GNNs and graph heterophily degree do not follow a strict monotonic correlation, necessitating adaptive graph filters to guarantee good generalization performance. Hence, we propose HeroFilter, a simple yet powerful GNN, which extracts information across the heterophily spectrum and combines salient representations through adaptive mixing. HeroFilter's superior performance achieves up to 9.2% accuracy improvement over leading baselines across homophilic and heterophilic graphs.

A fundamental challenge of bipartite graph representation learning is how to extract informative node embeddings. Self-Supervised Learning (SSL) is a promising paradigm to address this challenge. Most recent bipartite graph SSL methods are based on contrastive learning which learns embeddings by discriminating positive and negative node pairs. Contrastive learning usually requires a large number of negative node pairs, which could lead to computational burden and semantic errors. In this paper, we introduce a novel synergistic representation learning model (STERLING) to learn node embeddings without negative node pairs. STERLING preserves the unique local and global synergies in bipartite graphs. The local synergies are captured by maximizing the similarity of the inter-type and intra-type positive node pairs, and the global synergies are captured by maximizing the mutual information of co-clusters. Theoretical analysis demonstrates that STERLING could improve the connectivity between different node types in the embedding space. Extensive empirical evaluation on various benchmark datasets and tasks demonstrates the effectiveness of STERLING for extracting node embeddings.

Graph learning plays a pivotal role in many high-impact application domains. Despite significant advances in this field, there is currently no single solution that effectively handles (1) data heterogeneity, (2) long-range dependencies, (3) graph heterophily, and (4) scalability to large graphs, all at the same time. Classical graph neural networks (GNNs) and graph transformers (GTs) address some but not all of these issues, often suffering from limitations such as quadratic computation complexity and/or suboptimal generalization performance in realistic applications. This paper introduces UnifiedGT, a novel framework that systematically addresses all of these challenges by automatically providing a graph transformer architecture with multiple components via neural architecture search. UnifiedGT consists of five major components: (1) graph sampling, (2) structural prior injection, (3) graph attention, (4) local/global information mixing, and (5) type-specific feedforward networks (FFNs). This modular approach enables the efficient processing of large-scale graphs and effective management of heterogeneity and heterophily while capturing long-range dependencies. We demonstrate the versatility of UnifiedGT through comprehensive experiments on several benchmark datasets, revealing insights such as the efficacy of graph sampling for GTs, the importance of explicit graph structure injection via attention masking, and the synergistic effect of local/global information mixing via a combination of global attention with local message passing. Furthermore, we formulate these design choices into a search space, where an optimal combination can be discovered for a particular dataset via neural architecture search. Notably, UnifiedGT improves generalization performance on various graph datasets, outperforming state-of-the-art GT models by a margin of about 3.7% on average. The framework is available on Github (https://github.com/junhongmit/H2GB) and PyPI (https://pypi.org/project/H2GB/), and documentation can be found at https://junhongmit.github.io/H2GB/.

There have been tremendous efforts over the past decades dedicated to the generation of realistic graphs in a variety of domains, ranging from social networks to computer networks, from gene regulatory networks to online transaction networks. Despite the remarkable success, the vast majority of these works are unsupervised in nature and are typically trained to minimize the expected graph reconstruction loss, which would result in the representation disparity issue in the generated graphs, i.e., the protected groups (often minorities) contribute less to the objective and thus suffer from systematically higher errors. In this paper, we aim to tailor graph generation to downstream mining tasks by leveraging label information and user-preferred parity constraints. In particular, we start from the investigation of representation disparity in the context of graph generative models. To mitigate the disparity, we propose a fairness-aware graph generative model named Fairgen. Our model jointly trains a label-informed graph generation module and a fair representation learning module by progressively learning the behaviors of the protected and unprotected groups, from the ‘easy’ concepts to the ‘hard’ ones. In addition, we propose a generic context sampling strategy for graph generative models, which is proven to be capable of fairly capturing the contextual information of each group with a high probability. Experimental results on seven real-world data sets demonstrate that Fairgen (1) obtains performance on par with state-of-the-art graph generative models across nine network properties, (2) mitigates the representation disparity issues in the generated graphs, and (3) substantially boosts the model performance by up to 17% in downstream tasks via data augmentation.

We study both stream-based and pool-based active learning with neural network approximations. A recent line of works proposed bandit-based approaches that transformed active learning into a bandit problem, achieving both theoretical and empirical success. However, the performance and computational costs of these methods may be susceptible to the number of classes, denoted as $K$, due to this transformation. Therefore, this paper seeks to answer the question: "How can we mitigate the adverse impacts of $K$ while retaining the advantages of principled exploration and provable performance guarantees in active learning?" To tackle this challenge, we propose two algorithms based on the newly designed exploitation and exploration neural networks for stream-based and pool-based active learning. Subsequently, we provide theoretical performance guarantees for both algorithms in a non-parametric setting, demonstrating a slower error-growth rate concerning $K$ for the proposed approaches. We use extensive experiments to evaluate the proposed algorithms, which consistently outperform state-of-the-art baselines.

Fairness-aware graph neural networks (GNNs) have gained a surge of attention as they can reduce the bias of predictions on any demographic group (e.g., female) in graph-based applications. Although these methods greatly improve the algorithmic fairness of GNNs, the fairness can be easily corrupted by carefully designed adversarial attacks. In this paper, we investigate the problem of adversarial attacks on fairness of GNNs and propose G-FairAttack, a general framework for attacking various types of fairness-aware GNNs in terms of fairness with an unnoticeable effect on prediction utility. In addition, we propose a fast computation technique to reduce the time complexity of G-FairAttack. The experimental study demonstrates that G-FairAttack successfully corrupts the fairness of different types of GNNs while keeping the attack unnoticeable. Our study on fairness attacks sheds light on potential vulnerabilities in fairness-aware GNNs and guides further research on the robustness of GNNs in terms of fairness.

Class imbalance is prevalent in real-world node classification tasks and poses great challenges for graph learning models. Most existing studies are rooted in a class-rebalancing (CR) perspective and address class imbalance with class-wise reweighting or resampling. In this work, we approach the root cause of class-imbalance bias from an topological paradigm. Specifically, we theoretically reveal two **fundamental phenomena in the graph topology** that greatly exacerbate the predictive bias stemming from class imbalance. On this basis, we devise a lightweight topological augmentation framework BAT to mitigate the class-imbalance bias without class rebalancing. Being orthogonal to CR, BAT can function as an **efficient plug-and-play module** that can be seamlessly combined with and significantly boost existing CR techniques. Systematic experiments on real-world imbalanced graph learning tasks show that BAT can deliver up to 46.27% performance gain and up to 72.74% bias reduction over existing techniques. Code, examples, and documentations are available at https://github.com/ZhiningLiu1998/BAT.

Hierarchical time series forecasting requires not only prediction accuracy but also coherency, i.e., forecasts add up appropriately across the hierarchy. Recent literature has shown that reconciliation via projection outperforms prior methods such as top-down or bottom-up approaches. Unlike existing work that pre-specifies a projection matrix (e.g., orthogonal), we study the problem of learning the optimal oblique projection from data for coherent forecasting of hierarchical time series. In addition to the unbiasedness-preserving property, oblique projection implicitly accounts for the hierarchy structure and assigns different weights to individual time series, providing significant adaptability over orthogonal projection which treats base forecast errors equally. We examine two broad classes of projections, namely Euclidean projection and general oblique projections. We propose to model the reconciliation step as a learnable, structured, projection layer in the neural forecaster architecture. The proposed approach allows for the efficient learning of the optimal projection in an end-to-end framework where both the neural forecaster and the projection layer are learned simultaneously. An empirical evaluation of real-world hierarchical time series datasets demonstrates the superior performance of the proposed method over existing state-of-the-art approaches.

Data collected in the real world often encapsulates historical discrimination against disadvantaged groups and individuals. Existing fair machine learning (FairML) research has predominantly focused on mitigating discriminative bias in the model prediction, with far less effort dedicated towards exploring how to trace biases present in the data, despite its importance for the transparency and interpretability of FairML. To fill this gap, we investigate a novel research problem: discovering samples that reflect biases/prejudices from the training data. Grounding on the existing fairness notions, we lay out a sample bias criterion and propose practical algorithms for measuring and countering sample bias. The derived bias score provides intuitive sample-level attribution and explanation of historical bias in data. On this basis, we further design two FairML strategies via sample-bias-informed minimal data editing. They can mitigate both group and individual unfairness at the cost of minimal or zero predictive utility loss. Extensive experiments and analyses on multiple real-world datasets demonstrate the effectiveness of our methods in explaining and mitigating unfairness. Code is available at https://github.com/ZhiningLiu1998/AIM.

In the context of long-tail classification on graphs, the vast majority of existing work primarily revolves around the development of model debiasing strategies, intending to mitigate class imbalances and enhance the overall performance. Despite the notable success, there is very limited literature that provides a theoretical tool for characterizing the behaviors of long-tail classes in graphs and gaining insight into generalization performance in real-world scenarios. To bridge this gap, we propose a generalization bound for long-tail classification on graphs by formulating the problem in the fashion of multi-task learning, i.e., each task corresponds to the prediction of one particular class. Our theoretical results show that the generalization performance of long-tail classification is dominated by the overall loss range and the task complexity. Building upon the theoretical findings, we propose a novel generic framework HierTail for long-tail classification on graphs. In particular, we start with a hierarchical task grouping module that allows us to assign related tasks into hypertasks and thus control the complexity of the task space; then, we further design a balanced contrastive learning module to adaptively balance the gradients of both head and tail classes to control the loss range across all tasks in a unified fashion. Extensive experiments demonstrate the effectiveness of HierTail in characterizing long-tail classes on real graphs, which achieves up to 12.9% improvement over the leading baseline method in balanced accuracy.

Yue Zhou, Yada Zhu, Diego Antognini, Yoon Kim, Yang Zhang. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2024.

_Graph Neural Tangent Kernel_ (GNTK) fuses graph neural networks and graph kernels, simplifies the process of graph representation learning, interprets the training dynamics of graph neural networks, and serves various applications like protein identification, image segmentation, and social network analysis. In practice, graph data carries complex information among entities that inevitably evolves over time, and previous static graph neural tangent kernel methods may be stuck in the sub-optimal solution in terms of both effectiveness and efficiency. As a result, extending the advantage of GNTK to temporal graphs becomes a critical problem. To this end, we propose the temporal graph neural tangent kernel, which not only extends the simplicity and interpretation ability of GNTK to the temporal setting but also leads to rigorous temporal graph classification error bounds. Furthermore, we prove that when the input temporal graph grows over time in the number of nodes, our temporal graph neural tangent kernel will converge in the limit to the _graphon_ NTK value, which implies the transferability and robustness of the proposed kernel method, named **Temp**oral **G**raph **N**eural **T**angent **K**ernel with **G**raphon-**G**uaranteed or **Temp-G$^3$NTK**. In addition to the theoretical analysis, we also perform extensive experiments, not only demonstrating the superiority of Temp-G$^3$NTK in the temporal graph classification task, but also showing that Temp-G^3NTK can achieve very competitive performance in node-level tasks like node classification compared with various SOTA graph kernel and representation learning baselines. Our code is available at https://github.com/kthrn22/TempGNTK.

Contrastive learning is an effective unsupervised method in graph representation learning. The key component of contrastive learning lies in the construction of positive and negative samples. Previous methods usually utilize the proximity of nodes in the graph as the principle. Recently, the data-augmentation-based contrastive learning method has advanced to show great power in the visual domain, and some works have extended this method from images to graphs. However, unlike the data augmentation on images, the data augmentation on graphs is far less intuitive and it is much harder to provide high-quality contrastive samples, which leaves much space for improvement. In this work, by introducing an adversarial graph view for data augmentation, we propose a simple but effective method, Adversarial Graph Contrastive Learning ( ArieL ), to extract informative contrastive samples within reasonable constraints. We develop a new technique called information regularization for stable training and use subgraph sampling for scalability. We generalize our method from node-level contrastive learning to the graph level by treating each graph instance as a super-node. ArieL consistently outperforms the current graph contrastive learning methods for both node-level and graph-level classification tasks on real-world datasets. We further demonstrate that ArieL is more robust in the face of adversarial attacks.

Multivariate time series (MTS) imputation is a widely studied problem in recent years. Existing methods can be divided into two main groups, including (1) deep recurrent or generative models that primarily focus on time series features, and (2) graph neural networks (GNNs) based models that utilize the topological information from the inherent graph structure of MTS as relational inductive bias for imputation. Nevertheless, these methods either neglect topological information or assume the graph structure is fixed and accurately known. Thus, they fail to fully utilize the graph dynamics for precise imputation in more challenging MTS data such as networked time series (NTS), where the underlying graph is constantly changing and might have missing edges. In this paper, we propose a novel approach to overcome these limitations. First, we define the problem of imputation over NTS which contains missing values in both node time series features and graph structures. Then, we design a new model named PoGeVon which leverages variational autoencoder (VAE) to predict missing values over both node time series features and graph structures. In particular, we propose a new node position embedding based on random walk with restart (RWR) in the encoder with provable higher expressive power compared with message-passing based graph neural networks (GNNs). We further design a decoder with 3-stage predictions from the perspective of multi-task learning to impute missing values in both time series and graph structures reciprocally. Experiment results demonstrate the effectiveness of our model over baselines.

In the era of big data, we are often facing the challenge of data heterogeneity and the lack of label information simultaneously. In the financial domain (e.g., fraud detection), the heterogeneous data may include not only numerical data (e.g., total debt and yearly income), but also text and images (e.g., financial statement and invoice images). At the same time, the label information (e.g., fraud transactions) may be missing for building predictive models. To address these challenges, many state-of-the-art multi-view clustering methods have been proposed and achieved outstanding performance. However, these methods typically do not take into consideration the fairness aspect and are likely to generate biased results using sensitive information such as race and gender. Therefore, in this paper, we propose a fairness-aware multi-view clustering method named Fair-MVC. It incorporates the group fairness constraint into the soft membership assignment for each cluster to ensure that the fraction of different groups in each cluster is approximately identical to the entire data set. Meanwhile, we adopt the idea of both contrastive learning and non-contrastive learning and propose novel regularizers to handle heterogeneous data in complex scenarios with missing data or noisy features. Experimental results on real-world data sets demonstrate the effectiveness and efficiency of the proposed framework. We also derive insights regarding the relative performance of the proposed regularizers in various scenarios.

With the advent of big data across multiple high-impact applications, we are often facing the challenge of complex heterogeneity. The newly collected data usually consist of multiple modalities and are characterized with multiple labels, thus exhibiting the co-existence of multiple types of heterogeneity. Although state-of-the-art techniques are good at modeling the complex heterogeneity with sufficient label information, such label information can be quite expensive to obtain in real applications. Recently, researchers pay great attention to contrastive learning due to its prominent performance by utilizing rich unlabeled data. However, existing work on contrastive learning is not able to address the problem of false-negative pairs, i.e., some 'negative' pairs may have similar representations if they have the same label. To overcome the issues, in this paper, we propose a unified heterogeneous learning framework, which combines both the weighted unsupervised contrastive loss and the weighted supervised contrastive loss to model multiple types of heterogeneity. We first provide a theoretical analysis showing that the vanilla contrastive learning loss easily leads to the sub-optimal solution in the presence of false-negative pairs, whereas the proposed weighted loss could automatically adjust the weight based on the similarity of the learned representations to mitigate this issue. Experimental results on real-world data sets demonstrate the effectiveness and the efficiency of the proposed framework modeling multiple types of heterogeneity.

Co-evolving sequences are ubiquitous in a variety of applications, where different sequences are often inherently inter-connected with each other. We refer to such sequences, together with their inherent connections modeled as a structured network, as network of co-evolving sequences (NoCES). Typical NoCES applications include road traffic monitoring, company revenue prediction, motion capture, etc. To date, it remains a daunting challenge to accurately model NoCES due to the coupling between network structure and sequences. In this paper, we propose to modeling \pname\ with the aim of simultaneously capturing both the dynamics and the interplay between network structure and sequences. Specifically, we propose a joint learning framework to alternatively update the network representations and sequence representations as the sequences evolve over time. A unique feature of our framework lies in that it can deal with the case when there are co-evolving sequences on both network nodes and edges. Experimental evaluations on four real datasets demonstrate that the proposed approach (1) outperforms the existing competitors in terms of prediction accuracy, and (2) scales linearly w.r.t. the sequence length and the network size.

Contrastive learning is an effective unsupervised method in graph representation learning. Recently, the data augmentation based contrastive learning method has been extended from images to graphs. However, most prior works are directly adapted from the models designed for images. Unlike the data augmentation on images, the data augmentation on graphs is far less intuitive and much harder to provide high-quality contrastive samples, which are the key to the performance of contrastive learning models. This leaves much space for improvement over the existing graph contrastive learning frameworks. In this work, by introducing an adversarial graph view and an information regularizer, we propose a simple but effective method, Adversarial Graph Contrastive Learning (ArieL), to extract informative contrastive samples within a reasonable constraint. It consistently outperforms the current graph contrastive learning methods in the node classification task over various real-world datasets and further improves the robustness of graph contrastive learning.

For time series data, certain types of outliers are intrinsically more harmful for parameter estimation and future predictions than others, irrespective of their frequency. In this paper, for the first time, we study the characteristics of such outliers through the lens of the influence functional from robust statistics. In particular, we consider the input time series as a contaminated process, with the recurring outliers generated from an unknown contaminating process. Then we leverage the influence functional to understand the impact of the contaminating process on parameter estimation. The influence functional results in a multi-dimensional vector that measures the sensitivity of the predictive model to the contaminating process, which can be challenging to interpret especially for models with a large number of parameters. To this end, we further propose a comprehensive single-valued metric (the SIF) to measure outlier impacts on future predictions. It provides a quantitative measure regarding the outlier impacts, which can be used in a variety of scenarios, such as the evaluation of outlier detection methods, the creation of more harmful outliers, etc. The empirical results on multiple real data sets demonstrate the effectivenss of the proposed SIF metric.

Wei Zhang, Ziming Huang, Yada Zhu, Guangnan Ye, Xiaodong Cui, Fan Zhang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Co-evolving time series appears in a multitude of applications such as environmental monitoring, financial analysis, and smart transportation. This paper aims to address the following challenges, including (C1) how to incorporate explicit relationship networks of the time series; (C2) how to model the implicit relationship of the temporal dynamics. We propose a novel model called Network of Tensor Time Series (NeT3), which is comprised of two modules, including Tensor Graph Convolutional Network (TGCN) and Tensor Recurrent Neural Network (TRNN). TGCN tackles the first challenge by generalizing Graph Convolutional Network (GCN) for flat graphs to tensor graphs, which captures the synergy between multiple graphs associated with the tensors. TRNN leverages tensor decomposition to model the implicit relationships among co-evolving time series. The experimental results on five real-world datasets demonstrate the efficacy of the proposed method.

Encoding a large-scale network into a low-dimensional space is a fundamental step for various network analytic problems, such as node classification, link prediction, community detection, etc. Existing methods focus on learning the network representation from either the static graphs or time-aggregated graphs (e.g., time-evolving graphs). However, many real systems are not static or time-aggregated as the nodes and edges are timestamped and dynamically changing over time. For examples, in anti-money laundering analysis, cycles formed with time-ordered transactions might be red flags in online transaction networks; in novelty detection, a star-shaped structure appearing in a short burst might be an underlying hot topic in social networks. Existing embedding models might not be able to well preserve such fine-grained network dynamics due to the incapability of dealing with continuous-time and the negligence of fine-grained interactions. To bridge this gap, in this paper, we propose a fine-grained temporal network embedding framework named FiGTNE, which aims to learn a comprehensive network representation that preserves the rich and complex network context in the temporal network. In particular, we start from the notion of fine-grained temporal networks, where the temporal network can be represented as a series of timestamped nodes and edges. Then, we propose the time-reinforced random walk (TRRW) with a bi-level context sampling strategy to explore the essential structures and temporal contexts in temporal networks. Extensive experimental results on real graphs demonstrate the efficacy of our FiGTNE framework.

Portfolio management (PM) is a fundamental financial planning task that aims to achieve investment goals such as maximal profits or minimal risks. Its decision process involves continuous derivation of valuable information from various data sources and sequential decision optimization, which is a prospective research direction for reinforcement learning (RL). In this paper, we propose SARL, a novel State-Augmented RL framework for PM. Our framework aims to address two unique challenges in financial PM: (1) data heterogeneity – the collected information for each asset is usually diverse, noisy and imbalanced (e.g., news articles); and (2) environment uncertainty – the financial market is versatile and non-stationary. To incorporate heterogeneous data and enhance robustness against environment uncertainty, our SARL augments the asset information with their price movement prediction as additional states, where the prediction can be solely based on financial data (e.g., asset prices) or derived from alternative sources such as news. Experiments on two real-world datasets, (i) Bitcoin market and (ii) HighTech stock market with 7-year Reuters news articles, validate the effectiveness of SARL over existing PM approaches, both in terms of accumulated profits and risk-adjusted profits. Moreover, extensive simulations are conducted to demonstrate the importance of our proposed state augmentation, providing new insights and boosting performance significantly over standard RL-based PM method and other baselines.

Real-world applications often involve domain-specific and task-based performance objectives that are not captured by the standard machine learning losses, but are critical for decision making. A key challenge for direct integration of more meaningful domain and task-based evaluation criteria into an end-to-end gradient-based training process is the fact that often such performance objectives are not necessarily differentiable and may even require additional decision-making optimization processing. We propose the Task-Oriented Prediction Network (TOPNet), an end-to-end learning scheme that automatically integrates task-based evaluation criteria into the learning process via a learnable surrogate loss function, which directly guides the model towards the task-based goal. A major benefit of the proposed TOPNet learning scheme lies in its capability of automatically integrating non-differentiable evaluation criteria, which makes it particularly suitable for diversified and customized task-based evaluation criteria in real-world tasks. We validate the performance of TOPNet on two real-world financial prediction tasks, revenue surprise forecasting and credit risk modeling. The experimental results demonstrate that TOPNet significantly outperforms both traditional modeling with standard losses and modeling with hand-crafted heuristic differentiable surrogate losses.

Financial time series analysis plays a central role in optimizing investment decision and hedging market risks. This is a challenging task as the problems are always accompanied by dual-level (i.e, data-level and task-level) heterogeneity. For instance, in stock price forecasting, a successful portfolio with bounded risks usually consists of a large number of stocks from diverse domains (e.g, utility, information technology, healthcare, etc.), and forecasting stocks in each domain can be treated as one task; within a portfolio, each stock is characterized by temporal data collected from multiple modalities (e.g, finance, weather, and news), which corresponds to the data-level heterogeneity. Furthermore, the finance industry follows highly regulated processes, which require prediction models to be interpretable, and the output results to meet compliance. Therefore, a natural research question is how to build a model that can achieve satisfactory performance on such multi-modality multi-task learning problems, while being able to provide comprehensive explanations for the end users.

Deep neural network clustering is superior to the conventional clustering methods due to deep feature extraction and nonlinear dimensionality reduction. Nevertheless, deep neural network leads to a rough representation regarding the inherent relationship of the data points. Therefore, it is still difficult for deep neural network to exploit the effective structure for direct clustering. To address this issue, we propose a robust embedded deep K-means clustering (RED-KC) method. The proposed RED-KC approach utilizes the δ-norm metric to constrain the feature mapping process of the auto-encoder network, so that data are mapped to a latent feature space, which is more conducive to the robust clustering. Compared to the existing auto-encoder networks with the fixed prior, the proposed RED-KC is adaptive during the process of feature mapping. More importantly, the proposed RED-KC embeds the clustering process with the auto-encoder network, such that deep feature extraction and clustering can be performed simultaneously. Accordingly, a direct and efficient clustering could be obtained within only one step to avoid the inconvenience of multiple separate stages, namely, losing pivotal information and correlation. Consequently, extensive experiments are provided to validate the effectiveness of the proposed approach.

With the rapid growth of e-tail, the cost to handle returned online orders also increases significantly and has become a major challenge in the e-commerce industry. Accurate prediction of product returns allows e-tailers to prevent problematic transactions in advance. However, the limited existing work for modeling customer online shopping behaviors and predicting their return actions fail to integrate the rich information in the product purchase and return history (e.g., return history, purchase-no-return behavior, and customer/product similarity). Furthermore, the large-scale data sets involved in this problem, typically consisting of millions of customers and tens of thousands of products, also render existing methods inefficient and ineffective at predicting the product returns. To address these problems, in this paper, we propose to use a weighted hybrid graph to represent the rich information in the product purchase and return history, in order to predict product returns. The proposed graph consists of both customer nodes and product nodes, undirected edges reflecting customer return history and customer/product similarity based on their attributes, as well as directed edges discriminating purchase-no-return and no-purchase actions. Based on this representation, we study a random-walk-based local algorithm for predicting product return propensity for each customer, whose computational complexity depends only on the size of the output cluster rather than the entire graph. Such a property makes the proposed local algorithm particularly suitable for processing the large-scale data sets to predict product returns. To test the performance of the proposed techniques, we evaluate the graph model and algorithm on multiple e-commerce data sets, showing improved performance over state-of-the-art methods.

Recent decades have witnessed the rapid growth of E-commerce. In particular, E-tail has provided customers with great convenience by allowing them to purchase retail products anywhere without visiting the actual stores. A recent trend in E-tail is to allow free shipping and hassle-free returns to further attract online customers. However, a downside of such a customer-friendly policy is the rapidly increasing return rate as well as the associated costs of handling returned online orders. Therefore, it has become imperative to take proactive measures for reducing the return rate and the associated cost. Despite the large amount of data available from historical purchase and return records, up until now, the problem of E-tail product return prediction has not attracted much attention from the data mining community.

Many real-world applications are characterized by temporal data collected from multiple modalities, each sampled with a different resolution. Examples include manufacturing processes and financial market prediction. In these applications, an interesting observation is that within the same modality, we often have data from multiple views, thus naturally forming a 2-level hierarchy: with the multiple modalities on the top, and the multiple views at the bottom. For example, in aluminum smelting processes, the multiple modalities include power, noise, alumina feed, etc; and within the same modality such as power, the different views correspond to various voltage, current and resistance control signals and measured responses. For such applications, we aim to address the following challenge, i.e., how can we integrate such multi-modality multi-resolution data to effectively predict the targets of interest, such as bath temperature in aluminum smelting cell and the volatility in financial market. In this paper, for the first time, we simultaneously model the hierarchical data structure and the multi-resolution property via a novel framework named HiMuV. Different from existing work based on multiple views on a single level or a single resolution, the proposed framework is based on the key assumption that the information from different modalities is complementary, whereas the information within the same modality (across different views) is redundant in terms of predicting the targets of interest. Therefore, we introduce an optimization framework where the objective function contains both the prediction loss and a novel regularizer enforcing the consistency among different views within the same modality. To solve this optimization framework, we propose an iterative algorithm based on randomized block coordinate descent. Experimental results on synthetic data, benchmark data, and various real data sets from aluminum smelting processes, and stock market prediction demonstrate the effectiveness and efficiency of the proposed algorithm.

User behavior modeling is essential in computational advertisement, which builds users' profiles by tracking their online behaviors and then delivers the relevant ads according to each user's interests and needs. Accurate models will lead to higher targeting accuracy and thus improved advertising performance. Intuitively, similar users tend to have similar behaviors towards the displayed ads (e.g., impression, click, conversion). However, to the best of our knowledge, there is not much previous work that explicitly investigates such similarities of various types of user behaviors, and incorporates them into ad response targeting and prediction, largely due to the prohibitive scale of the problem.

Many complex real applications involve the collection of time series data with multiple modalities and of multiple resolutions. For example, in aluminum smelting processes, the recorded process variables typically reflect various aspects of these processes, such as pressure and temperature, and they are often obtained with different time resolutions, such as every 5 minutes and every day. How can we effectively leverage both the multi-modality property and the multi-resolution property of the data for the sake of more accurate prediction of key process indicators (e.g., the cell temperature of the aluminum smelting processes)? Different from existing techniques, which can only model the multi-modality property or the multi-resolution property, in this paper, for the first time, we propose to jointly model the two properties such that the prediction results are consistent across multiple modalities and multiple resolutions. To this end, we construct an optimization framework, which is based on a novel regularizer imposing such consistency. Then, we design an effective and efficient optimization algorithm based on randomized block coordinate descent. Its performance is evaluated on both synthetic and real data sets, outperforming state-of-the-art techniques.

Recent years have witnessed data explosion in semiconductor manufacturing due to advances in instrumentation and storage techniques. The large amount of data associated with process variables monitored over time form a rich reservoir of information, which can be used for a variety of purposes, such as anomaly detection, quality control, and fault diagnostics. In particular, following the same recipe for a certain Integrated Circuit device, multiple tools and chambers can be deployed for the production of this device, during which multiple time series can be collected, such as temperature, impedance, gas flow, electric bias, etc. These time series naturally fit into a two-dimensional array (matrix), i.e., each element in this array corresponds to a time series for one process variable from one chamber. To leverage the rich structural information in such temporal data, in this article, we propose a novel framework named C-Struts to simultaneously cluster on the two dimensions of this array. In this framework, we interpret the structural information as a set of constraints on the cluster membership, introduce an auxiliary probability distribution accordingly, and design an iterative algorithm to assign each time series to a certain cluster on each dimension. Furthermore, we establish the equivalence between C-Struts and a generic optimization problem, which is able to accommodate various distance functions. Extensive experiments on synthetic, benchmark, as well as manufacturing datasets demonstrate the effectiveness of the proposed method.

In many real world applications such as image annotation, gene function prediction, and insider threat detection, the data collected from heterogeneous sources often exhibit multiple types of heterogeneity, such as task heterogeneity, view heterogeneity, and label heterogeneity. To address this problem, we propose a Hierarchical Multi-Latent Space (HiMLS) learning framework to jointly model the triple types of heterogeneity. The basic idea is to learn a hierarchical multi-latent space by which we can simultaneously leverage the task relatedness, view consistency and the label correlations to improve the learning performance. We first propose a multi-latent space approach to model the complex heterogeneity, which is then used as a building block to stack up a multi-layer structure in order to learn the hierarchical multi-latent space. In such a way, we can gradually learn the more abstract concepts in the higher level. We present two instantiated models of the generalized framework using different divergence measures. The two-phase learning algorithms are used to train the multi-layer models. We drive the multiplicative update rules for pre-training and fine-tuning in each model, and prove the convergence and correctness of the update methods. The effectiveness of the proposed approach is verified on various data sets.

Big data analytics is the latest spotlight with all the glare of fame ranging from media coverage to booming startup companies to eye-catching merges and acquisitions. On the contrary, the $336 billion industry of semiconductor was seen as an “old-fashioned” business, with fading interests from the best and brightest among young graduates and engineers. How will modern big data analytics help the semiconductor industry walk through this transition? This paper answers this question via a number of practical but challenging problems arising from semiconductor manufacturing process. We show that many existing machine learning algorithms are not well positioned to solve these problems, and novel techniques involving temporal, structural and hierarchical properties need to be developed to solve these problems.

This paper targets the problem of cargo pricing optimization in the air cargo business. Given the features associated with a pair of origination and destination, how can we simultaneously predict both the optimal price for the bid stage and the outcome of the transaction (win rate) in the decision stage? In addition, it is often the case that the matrix representing pairs of originations and destinations has a block structure, i.e., the originations and destinations can be co-clustered such that the predictive models are similar within the same co-cluster, and exhibit significant variation among different co-clusters. How can we uncover the co-clusters of originations and destinations while constructing the dual predictive models for the two stages?

Recent years have witnessed data explosion in semiconductor manufacturing due to advances in instrumentation and storage techniques. In particular, following the same recipe for a certain IC device, multiple tools and chambers can be deployed for the production of this device, during which multiple time series can be collected, such as temperature, impedance, gas flow, electric bias, etc. These time series naturally fit into a two-dimensional array (matrix), i.e., Each element in this array corresponds to a time series for one process variable from one chamber. To leverage the rich structural information in such temporal data, in this paper, we propose a novel framework named C-Struts to simultaneously cluster on the two dimensions of this array. In this framework, we interpret the structural information as a set of constraints on the cluster membership, introduce an auxiliary probability distribution accordingly, and design an iterative algorithm to assign each time series to a certain cluster on each dimension. To the best of our knowledge, we are the first to address this problem. Extensive experiments on benchmark and manufacturing data sets demonstrate the effectiveness of the proposed method.

In many real applications, the input data are naturally expressed as tensors, such as virtual metrology in semiconductor manufacturing, face recognition and gait recognition in computer vision, etc. In this paper, we propose a general optimization framework for dealing with tensor inputs. Most existing methods for supervised tensor learning use only rank-one weight tensors in the linear model and cannot readily incorporate domain knowledge. In our framework, we obtain the weight tensor in a hierarchical way — we first approximate it by a low-rank tensor, and then estimate the low-rank approximation using the prior knowledge from various sources, e.g., different domain experts. This is motivated by wafer quality prediction in semiconductor manufacturing. Furthermore, we propose an effective algorithm named H-MOTE for solving this framework, which is guaranteed to converge. The time complexity of H-MOTE is linear with respect to the number of examples as well as the size of the weight tensor. Experimental results show the superiority of H-MOTE over state-of-the-art techniques on both synthetic and real data sets.

Many real problems of multi-task learning exhibit hierarchical task relatedness. In other words, the tasks are partitioned into multiple groups. Different tasks within the same group are related on the task-level, whereas different groups are related on the group-level. For example, in semiconductor manufacturing, the problem of wafer quality prediction can be considered as hierarchical multi-task learning, where each task corresponds to a single side of a chamber with side-level relatedness, and a group of tasks corresponds to a chamber of multiple sides with chamber-level relatedness. Motivated by this application, in this paper, we propose an optimization framework for hierarchical multi-task learning, which partitions all the input features into 2 sets based on their characteristics, and models task-level and group-level relatedness by imposing different constraints on the coefficient vectors of the 2 sets. This is different from existing work on task clustering where the goal is to uncover the grouping of tasks, the tasks do not exhibit group-level relatedness, and the input features are not discriminated in the prediction model to model task-level and group-level relatedness. To solve this framework, we propose the \hear\ algorithm based on block coordinate descent, and demonstrate its effectiveness on both synthetic and real data sets from domains of semiconductor manufacturing and document classification.

At the port-of-entry, containers are inspected through a specific sequence of sensor stations to detect the presence of radioactive materials, biological and chemical agents, and other illegal cargo. The inspection policy, which includes the sequence in which sensors are applied and the threshold levels used at the inspection stations, affects the probability of misclassifying a container as well as the cost and time spent in inspection. This work is an extension of a paper by Elsayed et al., which considers an inspection system operating with a Boolean decision function combining station results. In this paper, we present a multiobjective optimization approach to determine the optimal sensor arrangement and threshold levels, while considering cost and time. The total cost includes cost incurred by misclassification errors and the total expected cost of inspection, while the time represents the total expected time a container spends in the inspection system. Examples which apply the approach in various systems are presented.

This paper considers the problem of container inspection at a port-of-entry. Containers are inspected through a specific sequence to detect the presence of nuclear materials, biological and chemical agents, and other illegal shipments. The threshold levels of sensors at inspection stations of the port-of-entry affect the probabilities of incorrectly accepting or rejecting a container. In this paper, we present several optimization approaches for determining the optimum sensor threshold levels, while considering misclassification errors, total cost of inspection, and budget constraints. In contrast to previous work which determines threshold levels and sequence separately, we consider an integrated system and determine them simultaneously. Examples applying the approaches in different sensor arrangements are demonstrated.