people

Cold-start rating prediction is a fundamental problem in recommender systems that has been extensively studied. Many methods have been proposed that exploit explicit relations among existing data, such as collaborative filtering, social recommendations and heterogeneous information network, to alleviate the data insufficiency issue for cold-start users and items. However, the explicit relations constructed based on data between different entities may be unreliable and irrelevant, which limits the performance ceiling of a specific recommendation task. Motivated by this, in this paper, we propose a flexible framework dubbed heterogeneous interaction rating network (HIRE). HIRE does not solely rely on pre-defined interaction patterns or a manually constructed heterogeneous information network. Instead, we devise a Heterogeneous Interaction Module (HIM) to jointly model heterogeneous interactions and directly infer the important interactions via the observed data. In the experiments, we evaluate our framework under three cold-start settings on three real-world datasets. The experimental results show that HIRE outperforms other baselines by a large margin. Furthermore, we visualize the inferred interactions of HIRE to reveal the intuition behind our framework.

Geometric graphs are a special kind of graph with geometric features, which are vital to model many scientific problems. Unlike generic graphs, geometric graphs often exhibit physical symmetries of translations, rotations, and reflections, making them ineffectively processed by current Graph Neural Networks (GNNs). To address this issue, researchers proposed a variety of geometric GNNs equipped with invariant/equivariant properties to better characterize the geometry and topology of geometric graphs. Given the current progress in this field, it is imperative to conduct a comprehensive survey of data structures, models, and applications related to geometric GNNs. In this paper, based on the necessary but concise mathematical preliminaries, we formalize geometric graph as the data structure, on top of which we provide a unified view of existing models from the geometric message passing perspective. Additionally, we summarize the applications as well as the related datasets to facilitate later research for methodology development and experimental evaluation. We also discuss the challenges and future potential directions of geometric GNNs at the end of this survey.

Incorporating Euclidean symmetries (e.g. rotation equivariance) as inductive biases into graph neural networks has improved their generalization ability and data efficiency in unbounded physical dynamics modeling. However, in various scientific and engineering applications, the symmetries of dynamics are frequently discrete due to the boundary conditions. Thus, existing GNNs either over-look necessary symmetry, resulting in suboptimal representation ability, or impose excessive equivariance, which fails to generalize to unobserved symmetric dynamics. In this work, we propose a general Discrete Equivariant Graph Neural Network (DEGNN) that guarantees equivariance to a given discrete point group. Specifically, we show that such discrete equivariant message passing could be constructed by transforming geometric features into permutation-invariant embeddings. Through relaxing continuous equivariant constraints, DEGNN can employ more geometric feature combinations to approximate unobserved physical object interaction functions. Two implementation approaches of DEGNN are proposed based on ranking or pooling permutation-invariant functions. We apply DEGNN to various physical dynamics, ranging from particle, molecular, crowd to vehicle dynamics. In twenty scenarios, DEGNN significantly outperforms existing state-of-the-art approaches. Moreover, we show that DEGNN is data efficient, learning with less data, and can generalize across scenarios such as unobserved orientation.

Attributed community search (ACS) aims to identify subgraphs satisfying both structure cohesiveness and attribute homogeneity in attributed graphs, for a given query that contains query nodes and query attributes. Previously, algorithmic approaches deal with ACS in a two-stage paradigm, which suffer from structural inflexibility and attribute irrelevance. To overcome this problem, recently, learning-based approaches have been proposed to learn both structures and attributes simultaneously as a one-stage paradigm. However, these approaches train a transductive model which assumes the graph to infer unseen queries is as same as the graph used for training. That limits the generalization and adaptation of these approaches to different heterogeneous graphs.In this paper, we propose a new framework, Inductive Attributed Community Search, IACS, by inductive learning, which can be used to infer new queries for different communities/graphs. Specifically, IACS employs an encoder-decoder neural architecture to handle an ACS task at a time, where a task consists of a graph with only a few queries and corresponding ground-truth. We design a three-phase workflow, "training-adaptation-inference", which learns a shared model to absorb and induce prior effective common knowledge about ACS across different tasks. And the shared model can swiftly adapt to a new task with small number of ground-truth. We conduct substantial experiments in 7 real-world datasets to verify the effectiveness of IACS for CS/ACS. Our approach IACS achieves 28.97% and 25.60% improvements in F1-score on average in CS and ACS, respectively.

Single-cell proteomics sequencing technology sheds light on protein–protein interactions, posttranslational modifications and proteoform dynamics in the cell. However, the uncertainty estimation for peptide quantification, data missingness, batch effects and high noise hinder the analysis of single-cell proteomic data. It is important to solve this set of tangled problems together, but the existing methods tailored for single-cell transcriptomes cannot fully address this task. Here we propose a versatile framework designed for single-cell proteomics data analysis called scPROTEIN, which consists of peptide uncertainty estimation based on a multitask heteroscedastic regression model and cell embedding generation based on graph contrastive learning. scPROTEIN can estimate the uncertainty of peptide quantification, denoise protein data, remove batch effects and encode single-cell proteomic-specific embeddings in a unified framework. We demonstrate that scPROTEIN is efficient for cell clustering, batch correction, cell type annotation, clinical analysis and spatially resolved proteomic data exploration.

Combinatorial medication recommendation (CMR) is a fundamental task of healthcare, which offers opportunities for clinical physicians to provide more precise prescriptions for patients with intricate health conditions, particularly in the scenarios of long-term medical care. Previous research efforts have sought to extract meaningful information from electronic health records (EHRs) to facilitate combinatorial medication recommendations. Existing learning-based approaches further consider the chemical structures of medications, but ignore the textual medication descriptions in which the functionalities are clearly described. Furthermore, the textual knowledge derived from the EHRs of patients remains largely underutilized. To address these issues, we introduce the Natural Language-Assisted Multi-modal Medication Recommendation (NLAMMR), a multi-modal alignment framework designed to learn knowledge from the patient view and medication view jointly. Specifically, NLAMMR formulates CMR as an alignment problem from patient and medication modalities. In this vein, we employ pretrained language models (PLMs) to extract in-domain knowledge regarding patients and medications, serving as the foundational representation for both modalities. In the medication modality, we exploit both chemical structures and textual descriptions to create medication representations. In the patient modality, we generate the patient representations based on textual descriptions of diagnosis, procedure, and symptom. Extensive experiments conducted on three publicly accessible datasets demonstrate that NLAMMR achieves new state-of-the-art performance, with a notable average improvement of 4.72% in Jaccard score.

Rapid and accurate state of health (SOH) estimation of retired batteries is a crucial pretreatment for reuse and recycling. However, data-driven methods require exhaustive data curation under random SOH and state of charge (SOC) retirement conditions. Here, we show that the generative learning-assisted SOH estimation is promising in alleviating data scarcity and heterogeneity challenges, validated through a pulse injection dataset of 2700 retired lithium-ion battery samples, covering 3 cathode material types, 3 physical formats, 4 capacity designs, and 4 historical usages with 10 SOC levels. Using generated data, a regressor realizes accurate SOH estimations, with mean absolute percentage errors below 6% under unseen SOC. We predict that assuming uniform deployment of the proposed technique, this would save 4.9 billion USD in electricity costs and 35.8 billion kg CO2 emissions by mitigating data curation costs for a 2030 worldwide battery retirement scenario. This paper highlights exploiting limited data for exploring extended data space using generative methods, given data can be time-consuming, expensive, and polluting to retrieve for many estimation and predictive tasks.

Unsorted retired batteries with varied cathode materials hinder the adoption of direct recycling due to their cathode-specific nature. The surge in retired batteries necessitates precise sorting for effective direct recycling, but challenges arise from varying operational histories, diverse manufacturers, and data privacy concerns of recycling collaborators (data owners). Here we show, from a unique dataset of 130 lithium-ion batteries spanning 5 cathode materials and 7 manufacturers, a federated machine learning approach can classify these retired batteries without relying on past operational data, safeguarding the data privacy of recycling collaborators. By utilizing the features extracted from the end-of-life charge-discharge cycle, our model exhibits 1% and 3% cathode sorting errors under homogeneous and heterogeneous battery recycling settings respectively, attributed to our innovative Wasserstein-distance voting strategy. Economically, the proposed method underscores the value of precise battery sorting for a prosperous and sustainable recycling industry. This study heralds a new paradigm of using privacy-sensitive data from diverse sources, facilitating collaborative and privacy-respecting decision-making for distributed systems.

Recent awareness of privacy protection and compliance requirement resulted in a controversial view of recommendation system due to personal data usage. Therefore, privacy-protected recommendation emerges as a novel research direction. In this paper, we first formulate this problem as a vertical federated learning problem, i.e., features are vertically distributed over different departments. We study a contextual bandit learning problem for recommendation in the vertical federated setting. To this end, we carefully design a customized encryption scheme named orthogonal matrix-based mask mechanism (O3M). O3M mechanism, a tailored component for contextual bandits by carefully exploiting their shared structure, can ensure privacy protection while avoiding expensive conventional cryptographic techniques. We further apply the mechanism to two commonly-used bandit algorithms, LinUCB and LinTS, and instantiate two practical protocols for online recommendation. The proposed protocols can perfectly recover the service quality of centralized bandit algorithms while achieving a satisfactory runtime efficiency, which is theoretically proved and analysed in this paper. By conducting extensive experiments on both synthetic and real-world datasets, we show the superiority of the proposed method in terms of privacy protection and recommendation performance.

Proteins govern a wide range of biological systems. Evaluating the changes in protein properties upon protein mutation is a fundamental application of protein design, where modeling the 3D protein structure is a principal task for AI-driven computational approaches. Existing deep learning (DL) approaches represent the protein structure as a 3D geometric graph and simplify the graph modeling to different degrees, thereby failing to capture the low-level atom patterns and high-level amino acid patterns simultaneously. In addition, limited training samples with ground truth labels and protein structures further restrict the effectiveness of DL approaches. In this paper, we propose a new graph learning framework, Hierarchical Graph Invariant Network (HGIN), a fine-grained and data-efficient graph neural encoder for encoding protein structures and predicting the mutation effect on protein properties. For fine-grained modeling, HGIN hierarchically models the low-level interactions of atoms and the high-level interactions of amino acid residues by Graph Neural Networks. For data efficiency, HGIN preserves the invariant encoding for atom permutation and coordinate transformation, which is an intrinsic inductive bias of property prediction that bypasses data augmentations. We integrate HGIN into a Siamese network to predict the quantitative effect on protein properties upon mutations. Our approach outperforms 9 state-of-the-art approaches on 3 protein datasets. More inspiringly, when predicting the neutralizing ability of human antibodies against COVID-19 mutant viruses, HGIN achieves an absolute improvement of 0.23 regarding the Spearman coefficient.

Cold-start rating prediction is a fundamental problem in recommender systems that has been extensively studied. Many methods have been proposed that exploit explicit relations among existing data, such as collaborative filtering, social recommendations and heterogeneous information network, to alleviate the data insufficiency issue for cold-start users and items. However, the explicit relations constructed based on data between different entities may be unreliable and irrelevant, which limits the performance ceiling of a specific recommendation task. Motivated by this, in this paper, we propose a flexible framework dubbed heterogeneous interaction rating network (HIRE). HIRE does not solely rely on pre-defined interaction patterns or a manually constructed heterogeneous information network. Instead, we devise a Heterogeneous Interaction Module (HIM) to jointly model heterogeneous interactions and directly infer the important interactions via the observed data. In the experiments, we evaluate our framework under three cold-start settings on three real-world datasets. The experimental results show that HIRE outperforms other baselines by a large margin. Furthermore, we visualize the inferred interactions of HIRE to reveal the intuition behind our framework.

Geometric graphs are a special kind of graph with geometric features, which are vital to model many scientific problems. Unlike generic graphs, geometric graphs often exhibit physical symmetries of translations, rotations, and reflections, making them ineffectively processed by current Graph Neural Networks (GNNs). To address this issue, researchers proposed a variety of geometric GNNs equipped with invariant/equivariant properties to better characterize the geometry and topology of geometric graphs. Given the current progress in this field, it is imperative to conduct a comprehensive survey of data structures, models, and applications related to geometric GNNs. In this paper, based on the necessary but concise mathematical preliminaries, we formalize geometric graph as the data structure, on top of which we provide a unified view of existing models from the geometric message passing perspective. Additionally, we summarize the applications as well as the related datasets to facilitate later research for methodology development and experimental evaluation. We also discuss the challenges and future potential directions of geometric GNNs at the end of this survey.

Incorporating Euclidean symmetries (e.g. rotation equivariance) as inductive biases into graph neural networks has improved their generalization ability and data efficiency in unbounded physical dynamics modeling. However, in various scientific and engineering applications, the symmetries of dynamics are frequently discrete due to the boundary conditions. Thus, existing GNNs either over-look necessary symmetry, resulting in suboptimal representation ability, or impose excessive equivariance, which fails to generalize to unobserved symmetric dynamics. In this work, we propose a general Discrete Equivariant Graph Neural Network (DEGNN) that guarantees equivariance to a given discrete point group. Specifically, we show that such discrete equivariant message passing could be constructed by transforming geometric features into permutation-invariant embeddings. Through relaxing continuous equivariant constraints, DEGNN can employ more geometric feature combinations to approximate unobserved physical object interaction functions. Two implementation approaches of DEGNN are proposed based on ranking or pooling permutation-invariant functions. We apply DEGNN to various physical dynamics, ranging from particle, molecular, crowd to vehicle dynamics. In twenty scenarios, DEGNN significantly outperforms existing state-of-the-art approaches. Moreover, we show that DEGNN is data efficient, learning with less data, and can generalize across scenarios such as unobserved orientation.

Attributed community search (ACS) aims to identify subgraphs satisfying both structure cohesiveness and attribute homogeneity in attributed graphs, for a given query that contains query nodes and query attributes. Previously, algorithmic approaches deal with ACS in a two-stage paradigm, which suffer from structural inflexibility and attribute irrelevance. To overcome this problem, recently, learning-based approaches have been proposed to learn both structures and attributes simultaneously as a one-stage paradigm. However, these approaches train a transductive model which assumes the graph to infer unseen queries is as same as the graph used for training. That limits the generalization and adaptation of these approaches to different heterogeneous graphs.In this paper, we propose a new framework, Inductive Attributed Community Search, IACS, by inductive learning, which can be used to infer new queries for different communities/graphs. Specifically, IACS employs an encoder-decoder neural architecture to handle an ACS task at a time, where a task consists of a graph with only a few queries and corresponding ground-truth. We design a three-phase workflow, "training-adaptation-inference", which learns a shared model to absorb and induce prior effective common knowledge about ACS across different tasks. And the shared model can swiftly adapt to a new task with small number of ground-truth. We conduct substantial experiments in 7 real-world datasets to verify the effectiveness of IACS for CS/ACS. Our approach IACS achieves 28.97% and 25.60% improvements in F1-score on average in CS and ACS, respectively.

Single-cell proteomics sequencing technology sheds light on protein–protein interactions, posttranslational modifications and proteoform dynamics in the cell. However, the uncertainty estimation for peptide quantification, data missingness, batch effects and high noise hinder the analysis of single-cell proteomic data. It is important to solve this set of tangled problems together, but the existing methods tailored for single-cell transcriptomes cannot fully address this task. Here we propose a versatile framework designed for single-cell proteomics data analysis called scPROTEIN, which consists of peptide uncertainty estimation based on a multitask heteroscedastic regression model and cell embedding generation based on graph contrastive learning. scPROTEIN can estimate the uncertainty of peptide quantification, denoise protein data, remove batch effects and encode single-cell proteomic-specific embeddings in a unified framework. We demonstrate that scPROTEIN is efficient for cell clustering, batch correction, cell type annotation, clinical analysis and spatially resolved proteomic data exploration.

Combinatorial medication recommendation (CMR) is a fundamental task of healthcare, which offers opportunities for clinical physicians to provide more precise prescriptions for patients with intricate health conditions, particularly in the scenarios of long-term medical care. Previous research efforts have sought to extract meaningful information from electronic health records (EHRs) to facilitate combinatorial medication recommendations. Existing learning-based approaches further consider the chemical structures of medications, but ignore the textual medication descriptions in which the functionalities are clearly described. Furthermore, the textual knowledge derived from the EHRs of patients remains largely underutilized. To address these issues, we introduce the Natural Language-Assisted Multi-modal Medication Recommendation (NLAMMR), a multi-modal alignment framework designed to learn knowledge from the patient view and medication view jointly. Specifically, NLAMMR formulates CMR as an alignment problem from patient and medication modalities. In this vein, we employ pretrained language models (PLMs) to extract in-domain knowledge regarding patients and medications, serving as the foundational representation for both modalities. In the medication modality, we exploit both chemical structures and textual descriptions to create medication representations. In the patient modality, we generate the patient representations based on textual descriptions of diagnosis, procedure, and symptom. Extensive experiments conducted on three publicly accessible datasets demonstrate that NLAMMR achieves new state-of-the-art performance, with a notable average improvement of 4.72% in Jaccard score.

Rapid and accurate state of health (SOH) estimation of retired batteries is a crucial pretreatment for reuse and recycling. However, data-driven methods require exhaustive data curation under random SOH and state of charge (SOC) retirement conditions. Here, we show that the generative learning-assisted SOH estimation is promising in alleviating data scarcity and heterogeneity challenges, validated through a pulse injection dataset of 2700 retired lithium-ion battery samples, covering 3 cathode material types, 3 physical formats, 4 capacity designs, and 4 historical usages with 10 SOC levels. Using generated data, a regressor realizes accurate SOH estimations, with mean absolute percentage errors below 6% under unseen SOC. We predict that assuming uniform deployment of the proposed technique, this would save 4.9 billion USD in electricity costs and 35.8 billion kg CO2 emissions by mitigating data curation costs for a 2030 worldwide battery retirement scenario. This paper highlights exploiting limited data for exploring extended data space using generative methods, given data can be time-consuming, expensive, and polluting to retrieve for many estimation and predictive tasks.

Unsorted retired batteries with varied cathode materials hinder the adoption of direct recycling due to their cathode-specific nature. The surge in retired batteries necessitates precise sorting for effective direct recycling, but challenges arise from varying operational histories, diverse manufacturers, and data privacy concerns of recycling collaborators (data owners). Here we show, from a unique dataset of 130 lithium-ion batteries spanning 5 cathode materials and 7 manufacturers, a federated machine learning approach can classify these retired batteries without relying on past operational data, safeguarding the data privacy of recycling collaborators. By utilizing the features extracted from the end-of-life charge-discharge cycle, our model exhibits 1% and 3% cathode sorting errors under homogeneous and heterogeneous battery recycling settings respectively, attributed to our innovative Wasserstein-distance voting strategy. Economically, the proposed method underscores the value of precise battery sorting for a prosperous and sustainable recycling industry. This study heralds a new paradigm of using privacy-sensitive data from diverse sources, facilitating collaborative and privacy-respecting decision-making for distributed systems.

Recent awareness of privacy protection and compliance requirement resulted in a controversial view of recommendation system due to personal data usage. Therefore, privacy-protected recommendation emerges as a novel research direction. In this paper, we first formulate this problem as a vertical federated learning problem, i.e., features are vertically distributed over different departments. We study a contextual bandit learning problem for recommendation in the vertical federated setting. To this end, we carefully design a customized encryption scheme named orthogonal matrix-based mask mechanism (O3M). O3M mechanism, a tailored component for contextual bandits by carefully exploiting their shared structure, can ensure privacy protection while avoiding expensive conventional cryptographic techniques. We further apply the mechanism to two commonly-used bandit algorithms, LinUCB and LinTS, and instantiate two practical protocols for online recommendation. The proposed protocols can perfectly recover the service quality of centralized bandit algorithms while achieving a satisfactory runtime efficiency, which is theoretically proved and analysed in this paper. By conducting extensive experiments on both synthetic and real-world datasets, we show the superiority of the proposed method in terms of privacy protection and recommendation performance.

Proteins govern a wide range of biological systems. Evaluating the changes in protein properties upon protein mutation is a fundamental application of protein design, where modeling the 3D protein structure is a principal task for AI-driven computational approaches. Existing deep learning (DL) approaches represent the protein structure as a 3D geometric graph and simplify the graph modeling to different degrees, thereby failing to capture the low-level atom patterns and high-level amino acid patterns simultaneously. In addition, limited training samples with ground truth labels and protein structures further restrict the effectiveness of DL approaches. In this paper, we propose a new graph learning framework, Hierarchical Graph Invariant Network (HGIN), a fine-grained and data-efficient graph neural encoder for encoding protein structures and predicting the mutation effect on protein properties. For fine-grained modeling, HGIN hierarchically models the low-level interactions of atoms and the high-level interactions of amino acid residues by Graph Neural Networks. For data efficiency, HGIN preserves the invariant encoding for atom permutation and coordinate transformation, which is an intrinsic inductive bias of property prediction that bypasses data augmentations. We integrate HGIN into a Siamese network to predict the quantitative effect on protein properties upon mutations. Our approach outperforms 9 state-of-the-art approaches on 3 protein datasets. More inspiringly, when predicting the neutralizing ability of human antibodies against COVID-19 mutant viruses, HGIN achieves an absolute improvement of 0.23 regarding the Spearman coefficient.