Toyotaro Suzumura

Renhe Jiang, Zhaonan Wang, Jiawei Yong et al.

Traffic forecasting as a canonical task of multivariate time series forecasting has been a significant research topic in AI community. To address the spatio-temporal heterogeneity and non-stationarity implied in the traffic stream, in this study, we propose Spatio-Temporal Meta-Graph Learning as a novel Graph Structure Learning mechanism on spatio-temporal data. Specifically, we implement this idea into Meta-Graph Convolutional Recurrent Network (MegaCRN) by plugging the Meta-Graph Learner powered by a Meta-Node Bank into GCRN encoder-decoder. We conduct a comprehensive evaluation on two benchmark datasets (i.e., METR-LA and PEMS-BAY) and a new large-scale traffic speed dataset called EXPY-TKY that covers 1843 expressway road links in Tokyo. Our model outperformed the state-of-the-arts on all three datasets. Besides, through a series of qualitative evaluations, we demonstrate that our model can explicitly disentangle the road links and time slots with different patterns and be robustly adaptive to any anomalous traffic situations. Codes and datasets are available at https://github.com/deepkashiwa20/MegaCRN.

LLM-jp, Akiko Aizawa, Eiji Aramaki et al.

This paper introduces LLM-jp, a cross-organizational project for the research and development of Japanese large language models (LLMs). LLM-jp aims to develop open-source and strong Japanese LLMs, and as of this writing, more than 1,500 participants from academia and industry are working together for this purpose. This paper presents the background of the establishment of LLM-jp, summaries of its activities, and technical reports on the LLMs developed by LLM-jp. For the latest activities, visit https://llm-jp.nii.ac.jp/en/.

Xiaohang Xu, Toyotaro Suzumura, Jiawei Yong et al.

Next Point-of-Interest (POI) recommendation plays a crucial role in urban mobility applications. Recently, POI recommendation models based on Graph Neural Networks (GNN) have been extensively studied and achieved, however, the effective incorporation of both spatial and temporal information into such GNN-based models remains challenging. Extracting distinct fine-grained features unique to each piece of information is difficult since temporal information often includes spatial information, as users tend to visit nearby POIs. To address the challenge, we propose \textbf{\underline{Mob}}ility \textbf{\underline{G}}raph \textbf{\underline{T}}ransformer (MobGT) that enables us to fully leverage graphs to capture both the spatial and temporal features in users' mobility patterns. MobGT combines individual spatial and temporal graph encoders to capture unique features and global user-location relations. Additionally, it incorporates a mobility encoder based on Graph Transformer to extract higher-order information between POIs. To address the long-tailed problem in spatial-temporal data, MobGT introduces a novel loss function, Tail Loss. Experimental results demonstrate that MobGT outperforms state-of-the-art models on various datasets and metrics, achieving 24\% improvement on average. Our codes are available at \url{https://github.com/Yukayo/MobGT}.

Renhe Jiang, Zhaonan Wang, Jiawei Yong et al.

Spatio-temporal modeling as a canonical task of multivariate time series forecasting has been a significant research topic in AI community. To address the underlying heterogeneity and non-stationarity implied in the graph streams, in this study, we propose Spatio-Temporal Meta-Graph Learning as a novel Graph Structure Learning mechanism on spatio-temporal data. Specifically, we implement this idea into Meta-Graph Convolutional Recurrent Network (MegaCRN) by plugging the Meta-Graph Learner powered by a Meta-Node Bank into GCRN encoder-decoder. We conduct a comprehensive evaluation on two benchmark datasets (METR-LA and PEMS-BAY) and a large-scale spatio-temporal dataset that contains a variaty of non-stationary phenomena. Our model outperformed the state-of-the-arts to a large degree on all three datasets (over 27% MAE and 34% RMSE). Besides, through a series of qualitative evaluations, we demonstrate that our model can explicitly disentangle locations and time slots with different patterns and be robustly adaptive to different anomalous situations. Codes and datasets are available at https://github.com/deepkashiwa20/MegaCRN.

Boming Yang, Dairui Liu, Toyotaro Suzumura et al.

Precisely recommending candidate news articles to users has always been a core challenge for personalized news recommendation systems. Most recent works primarily focus on using advanced natural language processing techniques to extract semantic information from rich textual data, employing content-based methods derived from local historical news. However, this approach lacks a global perspective, failing to account for users' hidden motivations and behaviors beyond semantic information. To address this challenge, we propose a novel model called GLORY (Global-LOcal news Recommendation sYstem), which combines global representations learned from other users with local representations to enhance personalized recommendation systems. We accomplish this by constructing a Global-aware Historical News Encoder, which includes a global news graph and employs gated graph neural networks to enrich news representations, thereby fusing historical news representations by a historical news aggregator. Similarly, we extend this approach to a Global Candidate News Encoder, utilizing a global entity graph and a candidate news aggregator to enhance candidate news representation. Evaluation results on two public news datasets demonstrate that our method outperforms existing approaches. Furthermore, our model offers more diverse recommendations.

Toyotaro Suzumura, Akiyoshi Sugiki, Hiroyuki Takizawa et al.

The growing amount of data and advances in data science have created a need for a new kind of cloud platform that provides users with flexibility, strong security, and the ability to couple with supercomputers and edge devices through high-performance networks. We have built such a nation-wide cloud platform, called "mdx" to meet this need. The mdx platform's virtualization service, jointly operated by 9 national universities and 2 national research institutes in Japan, launched in 2021, and more features are in development. Currently mdx is used by researchers in a wide variety of domains, including materials informatics, geo-spatial information science, life science, astronomical science, economics, social science, and computer science. This paper provides an the overview of the mdx platform, details the motivation for its development, reports its current status, and outlines its future plans.

Hiroki Kanezashi, Toyotaro Suzumura, Xin Liu et al.

While transactions with cryptocurrencies such as Ethereum are becoming more prevalent, fraud and other criminal transactions are not uncommon. Graph analysis algorithms and machine learning techniques detect suspicious transactions that lead to phishing in large transaction networks. Many graph neural network (GNN) models have been proposed to apply deep learning techniques to graph structures. Although there is research on phishing detection using GNN models in the Ethereum transaction network, models that address the scale of the number of vertices and edges and the imbalance of labels have not yet been studied. In this paper, we compared the model performance of GNN models on the actual Ethereum transaction network dataset and phishing reported label data to exhaustively compare and verify which GNN models and hyperparameters produce the best accuracy. Specifically, we evaluated the model performance of representative homogeneous GNN models which consider single-type nodes and edges and heterogeneous GNN models which support different types of nodes and edges. We showed that heterogeneous models had better model performance than homogeneous models. In particular, the RGCN model achieved the best performance in the overall metrics.

Anson Bastos, Abhishek Nadgeri, Kuldeep Singh et al.

Learning on evolving(dynamic) graphs has caught the attention of researchers as static methods exhibit limited performance in this setting. The existing methods for dynamic graphs learn spatial features by local neighborhood aggregation, which essentially only captures the low pass signals and local interactions. In this work, we go beyond current approaches to incorporate global features for effectively learning representations of a dynamically evolving graph. We propose to do so by capturing the spectrum of the dynamic graph. Since static methods to learn the graph spectrum would not consider the history of the evolution of the spectrum as the graph evolves with time, we propose a novel approach to learn the graph wavelets to capture this evolving spectra. Further, we propose a framework that integrates the dynamically captured spectra in the form of these learnable wavelets into spatial features for incorporating local and global interactions. Experiments on eight standard datasets show that our method significantly outperforms related methods on various tasks for dynamic graphs.

Md Mostafizur Rahman, Daisuke Kikuta, Satyen Abrol et al.

Lookalike models are based on the assumption that user similarity plays an important role towards product selling and enhancing the existing advertising campaigns from a very large user base. Challenges associated to these models reside on the heterogeneity of the user base and its sparsity. In this work, we propose a novel framework that unifies the customers different behaviors or features such as demographics, buying behaviors on different platforms, customer loyalty behaviors and build a lookalike model to improve customer targeting for Rakuten Group, Inc. Extensive experiments on real e-commerce and travel datasets demonstrate the effectiveness of our proposed lookalike model for user targeting task.

Anson Bastos, Kuldeep Singh, Abhishek Nadgeri et al.

In this paper we present a novel method, $\textit{Knowledge Persistence}$ ($\mathcal{KP}$), for faster evaluation of Knowledge Graph (KG) completion approaches. Current ranking-based evaluation is quadratic in the size of the KG, leading to long evaluation times and consequently a high carbon footprint. $\mathcal{KP}$ addresses this by representing the topology of the KG completion methods through the lens of topological data analysis, concretely using persistent homology. The characteristics of persistent homology allow $\mathcal{KP}$ to evaluate the quality of the KG completion looking only at a fraction of the data. Experimental results on standard datasets show that the proposed metric is highly correlated with ranking metrics (Hits@N, MR, MRR). Performance evaluation shows that $\mathcal{KP}$ is computationally efficient: In some cases, the evaluation time (validation+test) of a KG completion method has been reduced from 18 hours (using Hits@10) to 27 seconds (using $\mathcal{KP}$), and on average (across methods & data) reduces the evaluation time (validation+test) by $\approx$ $\textbf{99.96}\%$.

Daisuke Kikuta, Toyotaro Suzumura, Md Mostafizur Rahman et al.

Leveraging graphs on recommender systems has gained popularity with the development of graph representation learning (GRL). In particular, knowledge graph embedding (KGE) and graph neural networks (GNNs) are representative GRL approaches, which have achieved the state-of-the-art performance on several recommendation tasks. Furthermore, combination of KGE and GNNs (KG-GNNs) has been explored and found effective in many academic literatures. One of the main characteristics of GNNs is their ability to retain structural properties among neighbors in the resulting dense representation, which is usually coined as smoothing. The smoothing is specially desired in the presence of homophilic graphs, such as the ones we find on recommender systems. In this paper, we propose a new model for recommender systems named Knowledge Query-based Graph Convolution (KQGC). In contrast to exisiting KG-GNNs, KQGC focuses on the smoothing, and leverages a simple linear graph convolution for smoothing KGE. A pre-trained KGE is fed into KQGC, and it is smoothed by aggregating neighbor knowledge queries, which allow entity-embeddings to be aligned on appropriate vector points for smoothing KGE effectively. We apply the proposed KQGC to a recommendation task that aims prospective users for specific products. Extensive experiments on a real E-commerce dataset demonstrate the effectiveness of KQGC.

Shun Takashige, Masatoshi Hanai, Toyotaro Suzumura et al.

The prediction of material properties plays a crucial role in the development and discovery of materials in diverse applications, such as batteries, semiconductors, catalysts, and pharmaceuticals. Recently, there has been a growing interest in employing data-driven approaches by using machine learning technologies, in combination with conventional theoretical calculations. In material science, the prediction of unobserved values, commonly referred to as extrapolation, is particularly critical for property prediction as it enables researchers to gain insight into materials beyond the limits of available data. However, even with the recent advancements in powerful machine learning models, accurate extrapolation is still widely recognized as a significantly challenging problem. On the other hand, self-supervised pretraining is a machine learning technique where a model is first trained on unlabeled data using relatively simple pretext tasks before being trained on labeled data for target tasks. As self-supervised pretraining can effectively utilize material data without observed property values, it has the potential to improve the model's extrapolation ability. In this paper, we clarify how such self-supervised pretraining can enhance extrapolation performance.We propose an experimental framework for the demonstration and empirically reveal that while models were unable to accurately extrapolate absolute property values, self-supervised pretraining enables them to learn relative tendencies of unobserved property values and improve extrapolation performance.

Igor L. R. Azevedo, Toyotaro Suzumura, Yuichiro Yasui

In recent years, journalists have expressed concerns about the increasing trend of news article avoidance, especially within specific domains. This issue has been exacerbated by the rise of recommender systems. Our research indicates that recommender systems should consider avoidance as a fundamental factor. We argue that news articles can be characterized by three principal elements: exposure, relevance, and avoidance, all of which are closely interconnected. To address these challenges, we introduce AWRS, an Avoidance-Aware Recommender System. This framework incorporates avoidance awareness when recommending news, based on the premise that news article avoidance conveys significant information about user preferences. Evaluation results on three news datasets in different languages (English, Norwegian, and Japanese) demonstrate that our method outperforms existing approaches.

Limin Wang, Masatoshi Hanai, Toyotaro Suzumura et al.

Revealing and analyzing the various properties of materials is an essential and critical issue in the development of materials, including batteries, semiconductors, catalysts, and pharmaceuticals. Traditionally, these properties have been determined through theoretical calculations and simulations. However, it is not practical to perform such calculations on every single candidate material. Recently, a combination method of the theoretical calculation and machine learning has emerged, that involves training machine learning models on a subset of theoretical calculation results to construct a surrogate model that can be applied to the remaining materials. On the other hand, a technique called pre-training is used to improve the accuracy of machine learning models. Pre-training involves training the model on pretext task, which is different from the target task, before training the model on the target task. This process aims to extract the input data features, stabilizing the learning process and improving its accuracy. However, in the case of molecular property prediction, there is a strong imbalance in the distribution of input data and features, which may lead to biased learning towards frequently occurring data during pre-training. In this study, we propose an effective pre-training method that addresses the imbalance in input data. We aim to improve the final accuracy by modifying the loss function of the existing representative pre-training method, node masking, to compensate the imbalance. We have investigated and assessed the impact of our proposed imbalance compensation on pre-training and the final prediction accuracy through experiments and evaluations using benchmark of molecular property prediction models.

Aldo Pareja, Giacomo Domeniconi, Jie Chen et al.

Graph representation learning resurges as a trending research subject owing to the widespread use of deep learning for Euclidean data, which inspire various creative designs of neural networks in the non-Euclidean domain, particularly graphs. With the success of these graph neural networks (GNN) in the static setting, we approach further practical scenarios where the graph dynamically evolves. Existing approaches typically resort to node embeddings and use a recurrent neural network (RNN, broadly speaking) to regulate the embeddings and learn the temporal dynamics. These methods require the knowledge of a node in the full time span (including both training and testing) and are less applicable to the frequent change of the node set. In some extreme scenarios, the node sets at different time steps may completely differ. To resolve this challenge, we propose EvolveGCN, which adapts the graph convolutional network (GCN) model along the temporal dimension without resorting to node embeddings. The proposed approach captures the dynamism of the graph sequence through using an RNN to evolve the GCN parameters. Two architectures are considered for the parameter evolution. We evaluate the proposed approach on tasks including link prediction, edge classification, and node classification. The experimental results indicate a generally higher performance of EvolveGCN compared with related approaches. The code is available at \url{https://github.com/IBM/EvolveGCN}.

Limin Wang, Toyotaro Suzumura, Hiroki Kanezashi

Electroencephalography (EEG) signals provide critical insights for applications in disease diagnosis and healthcare. However, the scarcity of labeled EEG data poses a significant challenge. Foundation models offer a promising solution by leveraging large-scale unlabeled data through pre-training, enabling strong performance across diverse tasks. While both temporal dynamics and inter-channel relationships are vital for understanding EEG signals, existing EEG foundation models primarily focus on the former, overlooking the latter. To address this limitation, we propose Graph-Enhanced EEG Foundation Model (GEFM), a novel foundation model for EEG that integrates both temporal and inter-channel information. Our architecture combines Graph Neural Networks (GNNs), which effectively capture relational structures, with a masked autoencoder to enable efficient pre-training. We evaluated our approach using three downstream tasks and experimented with various GNN architectures. The results demonstrate that our proposed model, particularly when employing the GCN architecture with optimized configurations, consistently outperformed baseline methods across all tasks. These findings suggest that our model serves as a robust foundation model for EEG analysis.

Anson Bastos, Kuldeep Singh, Abhishek Nadgeri et al.

We present the Evolving Graph Fourier Transform (EFT), the first invertible spectral transform that captures evolving representations on temporal graphs. We motivate our work by the inadequacy of existing methods for capturing the evolving graph spectra, which are also computationally expensive due to the temporal aspect along with the graph vertex domain. We view the problem as an optimization over the Laplacian of the continuous time dynamic graph. Additionally, we propose pseudo-spectrum relaxations that decompose the transformation process, making it highly computationally efficient. The EFT method adeptly captures the evolving graph's structural and positional properties, making it effective for downstream tasks on evolving graphs. Hence, as a reference implementation, we develop a simple neural model induced with EFT for capturing evolving graph spectra. We empirically validate our theoretical findings on a number of large-scale and standard temporal graph benchmarks and demonstrate that our model achieves state-of-the-art performance.

Yuji Tone, Masatoshi Hanai, Mitsuaki Kawamura et al.

The discovery of new materials using crystal structure prediction (CSP) based on generative machine learning models has become a significant research topic in recent years. In this paper, we study invariance and continuity in the generative machine learning for CSP. We propose a new model, called ContinuouSP, which effectively handles symmetry and periodicity in crystals. We clearly formulate the invariance and the continuity, and construct a model based on the energy-based model. Our preliminary evaluation demonstrates the effectiveness of this model with the CSP task.

Pablo Loyola, Kento Hasegawa, Andres Hoyos-Idobro et al.

While Annealing Machines (AM) have shown increasing capabilities in solving complex combinatorial problems, positioning themselves as a more immediate alternative to the expected advances of future fully quantum solutions, there are still scaling limitations. In parallel, Graph Neural Networks (GNN) have been recently adapted to solve combinatorial problems, showing competitive results and potentially high scalability due to their distributed nature. We propose a merging approach that aims at retaining both the accuracy exhibited by AMs and the representational flexibility and scalability of GNNs. Our model considers a compression step, followed by a supervised interaction where partial solutions obtained from the AM are used to guide local GNNs from where node feature representations are obtained and combined to initialize an additional GNN-based solver that handles the original graph's target problem. Intuitively, the AM can solve the combinatorial problem indirectly by infusing its knowledge into the GNN. Experiments on canonical optimization problems show that the idea is feasible, effectively allowing the AM to solve size problems beyond its original limits.

Toyotaro Suzumura, Hiroki Kanezashi, Shotaro Akahori

In diagnosing neurological disorders from electroencephalography (EEG) data, foundation models such as Transformers have been employed to capture temporal dynamics. Additionally, Graph Neural Networks (GNNs) are critical for representing the spatial relationships among EEG sensors. However, fine-tuning these large-scale models for both temporal and spatial features can be prohibitively large in computational cost, especially under the limited availability of labeled EEG datasets. We propose EEG-GraphAdapter (EGA), a parameter-efficient fine-tuning (PEFT) approach designed to address these challenges. EGA is integrated into a pre-trained temporal backbone model as a GNN-based module, freezing the backbone and allowing only the adapter to be fine-tuned. This enables the effective acquisition of EEG spatial representations, significantly reducing computational overhead and data requirements. Experimental evaluations on two healthcare-related downstream tasks-Major Depressive Disorder (MDD) and Abnormality Detection (TUAB)-show that EGA improves performance by up to 16.1% in F1-score compared with the backbone BENDR model, highlighting its potential for scalable and accurate EEG-based predictions.

Jonas le Fevre Sejersen, Toyotaro Suzumura, Erdal Kayacan

This paper presents a novel approach to multi-robot collision avoidance that integrates global path planning with local navigation strategies, utilizing attentive graph neural networks to manage dynamic interactions among agents. We introduce a local navigation model that leverages pre-planned global paths, allowing robots to adhere to optimal routes while dynamically adjusting to environmental changes. The models robustness is enhanced through the introduction of noise during training, resulting in superior performance in complex, dynamic environments. Our approach is evaluated against established baselines, including NH-ORCA, DRL-NAV, and GA3C-CADRL, across various structurally diverse simulated scenarios. The results demonstrate that our model achieves consistently higher success rates, lower collision rates, and more efficient navigation, particularly in challenging scenarios where baseline models struggle. This work offers an advancement in multi-robot navigation, with implications for robust performance in complex, dynamic environments with varying degrees of complexity, such as those encountered in logistics, where adaptability is essential for accommodating unforeseen obstacles and unpredictable changes.

Ziyin Xiao, Toyotaro Suzumura

Cross-Domain Recommender (CDR) systems aim to transfer knowledge from dense to sparse domains, alleviating data sparsity and cold-start issues in single-domain recommendation. While many methods assume overlapping users or items to connect domains, this is often unrealistic in real-world settings. Thus, non-overlapping CDR systems, which require no shared users or items, are needed. However, non-overlapping CDR is challenging due to: (1) the absence of overlap preventing direct bridges between domains, and (2) large distributional discrepancies degrading transfer performance. Moreover, most recommenders represent user preferences as discrete vectors, failing to capture their fine-grained, multi-faceted nature. We propose DUP-OT (Distributional User Preferences with Optimal Transport), a framework for non-overlapping CDR. DUP-OT has three stages: (1) Shared Preprocessing, where review-based embeddings and an autoencoder encode users and items from both domains; (2) User GMM Weight Learning, which models user preferences as Gaussian mixtures with learned weights; and (3) Cross-domain Rating Prediction, where optimal transport aligns Gaussian components across domains, enabling preference transfer from source to target. Experiments on Amazon review datasets show that DUP-OT effectively mitigates domain discrepancy and outperforms state-of-the-art baselines under the non-overlapping CDR setting.

Igor L. R. Azevedo, Toyotaro Suzumura

Stock market forecasting has been a topic of extensive research, aiming to provide investors with optimal stock recommendations for higher returns. In recent years, this field has gained even more attention due to the widespread adoption of deep learning models. While these models have achieved impressive accuracy in predicting stock behavior, tailoring them to specific scenarios has become increasingly important. Election Day represents one such critical scenario, characterized by intensified market volatility, as the winning candidate's policies significantly impact various economic sectors and companies. To address this challenge, we propose the Election Day Stock Market Forecasting (EDSMF) Model. Our approach leverages the contextual capabilities of large language models alongside specialized agents designed to analyze the political and economic consequences of elections. By building on a state-of-the-art architecture, we demonstrate that EDSMF improves the predictive performance of the S&P 500 during this uniquely volatile day.

Christopher Wolters, Xiaoxuan Yang, Ulf Schlichtmann et al.

Large language models (LLMs) have recently transformed natural language processing, enabling machines to generate human-like text and engage in meaningful conversations. This development necessitates speed, efficiency, and accessibility in LLM inference as the computational and memory requirements of these systems grow exponentially. Meanwhile, advancements in computing and memory capabilities are lagging behind, exacerbated by the discontinuation of Moore's law. With LLMs exceeding the capacity of single GPUs, they require complex, expert-level configurations for parallel processing. Memory accesses become significantly more expensive than computation, posing a challenge for efficient scaling, known as the memory wall. Here, compute-in-memory (CIM) technologies offer a promising solution for accelerating AI inference by directly performing analog computations in memory, potentially reducing latency and power consumption. By closely integrating memory and compute elements, CIM eliminates the von Neumann bottleneck, reducing data movement and improving energy efficiency. This survey paper provides an overview and analysis of transformer-based models, reviewing various CIM architectures and exploring how they can address the imminent challenges of modern AI computing systems. We discuss transformer-related operators and their hardware acceleration schemes and highlight challenges, trends, and insights in corresponding CIM designs.

Anson Bastos, Abhishek Nadgeri, Kuldeep Singh et al.

The recent works proposing transformer-based models for graphs have proven the inadequacy of Vanilla Transformer for graph representation learning. To understand this inadequacy, there is a need to investigate if spectral analysis of the transformer will reveal insights into its expressive power. Similar studies already established that spectral analysis of Graph neural networks (GNNs) provides extra perspectives on their expressiveness. In this work, we systematically study and establish the link between the spatial and spectral domain in the realm of the transformer. We further provide a theoretical analysis and prove that the spatial attention mechanism in the transformer cannot effectively capture the desired frequency response, thus, inherently limiting its expressiveness in spectral space. Therefore, we propose FeTA, a framework that aims to perform attention over the entire graph spectrum (i.e., actual frequency components of the graphs) analogous to the attention in spatial space. Empirical results suggest that FeTA provides homogeneous performance gain against vanilla transformer across all tasks on standard benchmarks and can easily be extended to GNN-based models with low-pass characteristics (e.g., GAT).

Venkatesan T. Chakaravarthy, Shivmaran S. Pandian, Saurabh Raje et al.

We present distributed algorithms for training dynamic Graph Neural Networks (GNN) on large scale graphs spanning multi-node, multi-GPU systems. To the best of our knowledge, this is the first scaling study on dynamic GNN. We devise mechanisms for reducing the GPU memory usage and identify two execution time bottlenecks: CPU-GPU data transfer; and communication volume. Exploiting properties of dynamic graphs, we design a graph difference-based strategy to significantly reduce the transfer time. We develop a simple, but effective data distribution technique under which the communication volume remains fixed and linear in the input size, for any number of GPUs. Our experiments using billion-size graphs on a system of 128 GPUs shows that: (i) the distribution scheme achieves up to 30x speedup on 128 GPUs; (ii) the graph-difference technique reduces the transfer time by a factor of up to 4.1x and the overall execution time by up to 40%

Irene Li, Aosong Feng, Hao Wu et al.

Multi-label text classification (MLTC) is an attractive and challenging task in natural language processing (NLP). Compared with single-label text classification, MLTC has a wider range of applications in practice. In this paper, we propose a label-interpretable graph convolutional network model to solve the MLTC problem by modeling tokens and labels as nodes in a heterogeneous graph. In this way, we are able to take into account multiple relationships including token-level relationships. Besides, the model allows better interpretability for predicted labels as the token-label edges are exposed. We evaluate our method on four real-world datasets and it achieves competitive scores against selected baseline methods. Specifically, this model achieves a gain of 0.14 on the F1 score in the small label set MLTC, and 0.07 in the large label set scenario.

Toyotaro Suzumura, Dario Garcia-Gasulla, Sergio Alvarez Napagao et al.

This paper aims at providing the summary of the Global Data Science Project (GDSC) for COVID-19. as on May 31 2020. COVID-19 has largely impacted on our societies through both direct and indirect effects transmitted by the policy measures to counter the spread of viruses. We quantitatively analysed the multifaceted impacts of the COVID-19 pandemic on our societies including people's mobility, health, and social behaviour changes. People's mobility has changed significantly due to the implementation of travel restriction and quarantine measurements. Indeed, the physical distance has widened at international (cross-border), national and regional level. At international level, due to the travel restrictions, the number of international flights has plunged overall at around 88 percent during March. In particular, the number of flights connecting Europe dropped drastically in mid of March after the United States announced travel restrictions to Europe and the EU and participating countries agreed to close borders, at 84 percent decline compared to March 10th. Similarly, we examined the impacts of quarantine measures in the major city: Tokyo (Japan), New York City (the United States), and Barcelona (Spain). Within all three cities, we found the significant decline in traffic volume. We also identified the increased concern for mental health through the analysis of posts on social networking services such as Twitter and Instagram. Notably, in the beginning of April 2020, the number of post with #depression on Instagram doubled, which might reflect the rise in mental health awareness among Instagram users. Besides, we identified the changes in a wide range of people's social behaviors, as well as economic impacts through the analysis of Instagram data and primary survey data.

Toyotaro Suzumura, Hiroki Kanezashi, Mishal Dholakia et al.

As COVID-19 transmissions spread worldwide, governments have announced and enforced travel restrictions to prevent further infections. Such restrictions have a direct effect on the volume of international flights among these countries, resulting in extensive social and economic costs. To better understand the situation in a quantitative manner, we used the Opensky network data to clarify flight patterns and flight densities around the world and observe relationships between flight numbers with new infections, and with the economy (unemployment rate) in Barcelona. We found that the number of daily flights gradually decreased and suddenly dropped 64% during the second half of March in 2020 after the US and Europe enacted travel restrictions. We also observed a 51% decrease in the global flight network density decreased during this period. Regarding new COVID-19 cases, the world had an unexpected surge regardless of travel restrictions. Finally, the layoffs for temporary workers in the tourism and airplane business increased by 4.3 fold in the weeks following Spain's decision to close its borders.

Irene Li, Yixin Li, Tianxiao Li et al.

The outbreak of coronavirus disease 2019 (COVID-19) recently has affected human life to a great extent. Besides direct physical and economic threats, the pandemic also indirectly impact people's mental health conditions, which can be overwhelming but difficult to measure. The problem may come from various reasons such as unemployment status, stay-at-home policy, fear for the virus, and so forth. In this work, we focus on applying natural language processing (NLP) techniques to analyze tweets in terms of mental health. We trained deep models that classify each tweet into the following emotions: anger, anticipation, disgust, fear, joy, sadness, surprise and trust. We build the EmoCT (Emotion-Covid19-Tweet) dataset for the training purpose by manually labeling 1,000 English tweets. Furthermore, we propose and compare two methods to find out the reasons that are causing sadness and fear.

Lucia Larise Stavarache, Donatas Narbutis, Toyotaro Suzumura et al.

In the recent years money laundering schemes have grown in complexity and speed of realization, affecting financial institutions and millions of customers globally. Strengthened privacy policies, along with in-country regulations, make it hard for banks to inner- and cross-share, and report suspicious activities for the AML (Anti-Money Laundering) measures. Existing topologies and models for AML analysis and information sharing are subject to major limitations, such as compliance with regulatory constraints, extended infrastructure to run high-computation algorithms, data quality and span, proving cumbersome and costly to execute, federate, and interpret. This paper proposes a new topology for exploring multi-banking customer social relations in AML context -- customer-to-customer, customer-to-transaction, and transaction-to-transaction -- using a 3D modeling topological algebra formulated through Poincaré embeddings.

Daiki Matsunaga, Toyotaro Suzumura, Toshihiro Takahashi

Recently, there has been a surge of interest in the use of machine learning to help aid in the accurate predictions of financial markets. Despite the exciting advances in this cross-section of finance and AI, many of the current approaches are limited to using technical analysis to capture historical trends of each stock price and thus limited to certain experimental setups to obtain good prediction results. On the other hand, professional investors additionally use their rich knowledge of inter-market and inter-company relations to map the connectivity of companies and events, and use this map to make better market predictions. For instance, they would predict the movement of a certain company's stock price based not only on its former stock price trends but also on the performance of its suppliers or customers, the overall industry, macroeconomic factors and trade policies. This paper investigates the effectiveness of work at the intersection of market predictions and graph neural networks, which hold the potential to mimic the ways in which investors make decisions by incorporating company knowledge graphs directly into the predictive model. The main goal of this work is to test the validity of this approach across different markets and longer time horizons for backtesting using rolling window analysis. In this work, we concentrate on the prediction of individual stock prices in the Japanese Nikkei 225 market over a period of roughly 20 years. For the knowledge graph, we use the Nikkei Value Search data, which is a rich dataset showing mainly supplier relations among Japanese and foreign companies. Our preliminary results show a 29.5% increase and a 2.2-fold increase in the return ratio and Sharpe ratio, respectively, when compared to the market benchmark, as well as a 6.32% increase and 1.3-fold increase, respectively, compared to the baseline LSTM model.

Toyotaro Suzumura, Yi Zhou, Natahalie Baracaldo et al.

Financial crime is a large and growing problem, in some way touching almost every financial institution. Financial institutions are the front line in the war against financial crime and accordingly, must devote substantial human and technology resources to this effort. Current processes to detect financial misconduct have limitations in their ability to effectively differentiate between malicious behavior and ordinary financial activity. These limitations tend to result in gross over-reporting of suspicious activity that necessitate time-intensive and costly manual review. Advances in technology used in this domain, including machine learning based approaches, can improve upon the effectiveness of financial institutions' existing processes, however, a key challenge that most financial institutions continue to face is that they address financial crimes in isolation without any insight from other firms. Where financial institutions address financial crimes through the lens of their own firm, perpetrators may devise sophisticated strategies that may span across institutions and geographies. Financial institutions continue to work relentlessly to advance their capabilities, forming partnerships across institutions to share insights, patterns and capabilities. These public-private partnerships are subject to stringent regulatory and data privacy requirements, thereby making it difficult to rely on traditional technology solutions. In this paper, we propose a methodology to share key information across institutions by using a federated graph learning platform that enables us to build more accurate machine learning models by leveraging federated learning and also graph learning approaches. We demonstrated that our federated model outperforms local model by 20% with the UK FCA TechSprint data set. This new platform opens up a door to efficiently detecting global money laundering activity.

Mark Weber, Jie Chen, Toyotaro Suzumura et al.

Organized crime inflicts human suffering on a genocidal scale: the Mexican drug cartels have murdered 150,000 people since 2006, upwards of 700,000 people per year are "exported" in a human trafficking industry enslaving an estimated 40 million people. These nefarious industries rely on sophisticated money laundering schemes to operate. Despite tremendous resources dedicated to anti-money laundering (AML) only a tiny fraction of illicit activity is prevented. The research community can help. In this brief paper, we map the structural and behavioral dynamics driving the technical challenge. We review AML methods, current and emergent. We provide a first look at scalable graph convolutional neural networks for forensic analysis of financial data, which is massive, dense, and dynamic. We report preliminary experimental results using a large synthetic graph (1M nodes, 9M edges) generated by a data simulator we created called AMLSim. We consider opportunities for high performance efficiency, in terms of computation and memory, and we share results from a simple graph compression experiment. Our results support our working hypothesis that graph deep learning for AML bears great promise in the fight against criminal financial activity.

Weiyi Liu, Zhining Liu, Toyotaro Suzumura et al.

Adding attributes for nodes to network embedding helps to improve the ability of the learned joint representation to depict features from topology and attributes simultaneously. Recent research on the joint embedding has exhibited a promising performance on a variety of tasks by jointly embedding the two spaces. However, due to the indispensable requirement of globality based information, present approaches contain a flaw of in-scalability. Here we propose \emph{SANE}, a scalable attribute-aware network embedding algorithm with locality, to learn the joint representation from topology and attributes. By enforcing the alignment of a local linear relationship between each node and its K-nearest neighbors in topology and attribute space, the joint embedding representations are more informative comparing with a single representation from topology or attributes alone. And we argue that the locality in \emph{SANE} is the key to learning the joint representation at scale. By using several real-world networks from diverse domains, We demonstrate the efficacy of \emph{SANE} in performance and scalability aspect. Overall, for performance on label classification, SANE successfully reaches up to the highest F1-score on most datasets, and even closer to the baseline method that needs label information as extra inputs, compared with other state-of-the-art joint representation algorithms. What's more, \emph{SANE} has an up to 71.4\% performance gain compared with the single topology-based algorithm. For scalability, we have demonstrated the linearly time complexity of \emph{SANE}. In addition, we intuitively observe that when the network size scales to 100,000 nodes, the "learning joint embedding" step of \emph{SANE} only takes $\approx10$ seconds.

Weiyi Liu, Hal Cooper, Min Hwan Oh et al.

Inspired by the generation power of generative adversarial networks (GANs) in image domains, we introduce a novel hierarchical architecture for learning characteristic topological features from a single arbitrary input graph via GANs. The hierarchical architecture consisting of multiple GANs preserves both local and global topological features and automatically partitions the input graph into representative stages for feature learning. The stages facilitate reconstruction and can be used as indicators of the importance of the associated topological structures. Experiments show that our method produces subgraphs retaining a wide range of topological features, even in early reconstruction stages (unlike a single GAN, which cannot easily identify such features, let alone reconstruct the original graph). This paper is firstline research on combining the use of GANs and graph topological analysis.

Armand Vilalta, Dario Garcia-Gasulla, Ferran Parés et al.

The current state-of-the-art for image annotation and image retrieval tasks is obtained through deep neural networks, which combine an image representation and a text representation into a shared embedding space. In this paper we evaluate the impact of using the Full-Network embedding in this setting, replacing the original image representation in a competitive multimodal embedding generation scheme. Unlike the one-layer image embeddings typically used by most approaches, the Full-Network embedding provides a multi-scale representation of images, which results in richer characterizations. To measure the influence of the Full-Network embedding, we evaluate its performance on three different datasets, and compare the results with the original multimodal embedding generation scheme when using a one-layer image embedding, and with the rest of the state-of-the-art. Results for image annotation and image retrieval tasks indicate that the Full-Network embedding is consistently superior to the one-layer embedding. These results motivate the integration of the Full-Network embedding on any multimodal embedding generation scheme, something feasible thanks to the flexibility of the approach.

Dario Garcia-Gasulla, Armand Vilalta, Ferran Parés et al.

Patterns stored within pre-trained deep neural networks compose large and powerful descriptive languages that can be used for many different purposes. Typically, deep network representations are implemented within vector embedding spaces, which enables the use of traditional machine learning algorithms on top of them. In this short paper we propose the construction of a graph embedding space instead, introducing a methodology to transform the knowledge coded within a deep convolutional network into a topological space (i.e. a network). We outline how such graph can hold data instances, data features, relations between instances and features, and relations among features. Finally, we introduce some preliminary experiments to illustrate how the resultant graph embedding space can be exploited through graph analytics algorithms.

Weiyi Liu, Pin-Yu Chen, Hal Cooper et al.

This paper is first-line research expanding GANs into graph topology analysis. By leveraging the hierarchical connectivity structure of a graph, we have demonstrated that generative adversarial networks (GANs) can successfully capture topological features of any arbitrary graph, and rank edge sets by different stages according to their contribution to topology reconstruction. Moreover, in addition to acting as an indicator of graph reconstruction, we find that these stages can also preserve important topological features in a graph.

Dario Garcia-Gasulla, Armand Vilalta, Ferran Parés et al.

Transfer learning for feature extraction can be used to exploit deep representations in contexts where there is very few training data, where there are limited computational resources, or when tuning the hyper-parameters needed for training is not an option. While previous contributions to feature extraction propose embeddings based on a single layer of the network, in this paper we propose a full-network embedding which successfully integrates convolutional and fully connected features, coming from all layers of a deep convolutional neural network. To do so, the embedding normalizes features in the context of the problem, and discretizes their values to reduce noise and regularize the embedding space. Significantly, this also reduces the computational cost of processing the resultant representations. The proposed method is shown to outperform single layer embeddings on several image classification tasks, while also being more robust to the choice of the pre-trained model used for obtaining the initial features. The performance gap in classification accuracy between thoroughly tuned solutions and the full-network embedding is also reduced, which makes of the proposed approach a competitive solution for a large set of applications.

Mark Hughes, Irene Li, Spyros Kotoulas et al.

We present an approach to automatically classify clinical text at a sentence level. We are using deep convolutional neural networks to represent complex features. We train the network on a dataset providing a broad categorization of health information. Through a detailed evaluation, we demonstrate that our method outperforms several approaches widely used in natural language processing tasks by about 15%.

Dario Garcia-Gasulla, Ferran Parés, Armand Vilalta et al.

Deep neural networks are representation learning techniques. During training, a deep net is capable of generating a descriptive language of unprecedented size and detail in machine learning. Extracting the descriptive language coded within a trained CNN model (in the case of image data), and reusing it for other purposes is a field of interest, as it provides access to the visual descriptors previously learnt by the CNN after processing millions of images, without requiring an expensive training phase. Contributions to this field (commonly known as feature representation transfer or transfer learning) have been purely empirical so far, extracting all CNN features from a single layer close to the output and testing their performance by feeding them to a classifier. This approach has provided consistent results, although its relevance is limited to classification tasks. In a completely different approach, in this paper we statistically measure the discriminative power of every single feature found within a deep CNN, when used for characterizing every class of 11 datasets. We seek to provide new insights into the behavior of CNN features, particularly the ones from convolutional layers, as this can be relevant for their application to knowledge representation and reasoning. Our results confirm that low and middle level features may behave differently to high level features, but only under certain conditions. We find that all CNN features can be used for knowledge representation purposes both by their presence or by their absence, doubling the information a single CNN feature may provide. We also study how much noise these features may include, and propose a thresholding approach to discard most of it. All these insights have a direct application to the generation of CNN embedding spaces.

Dario Garcia-Gasulla, Eduard Ayguadé, Jesús Labarta et al.

The hyperlink prediction task, that of proposing new links between webpages, can be used to improve search engines, expand the visibility of web pages, and increase the connectivity and navigability of the web. Hyperlink prediction is typically performed on webgraphs composed by thousands or millions of vertices, where on average each webpage contains less than fifty links. Algorithms processing graphs so large and sparse require to be both scalable and precise, a challenging combination. Similarity-based algorithms are among the most scalable solutions within the link prediction field, due to their parallel nature and computational simplicity. These algorithms independently explore the nearby topological features of every missing link from the graph in order to determine its likelihood. Unfortunately, the precision of similarity-based algorithms is limited, which has prevented their broad application so far. In this work we explore the performance of similarity-based algorithms for the particular problem of hyperlink prediction on large webgraphs, and propose a novel method which assumes the existence of hierarchical properties. We evaluate this new approach on several webgraphs and compare its performance with that of the current best similarity-based algorithms. Its remarkable performance leads us to argue on the applicability of the proposal, identifying several use cases of hyperlink prediction. We also describes the approach we took for the computation of large-scale graphs from the perspective of high-performance computing, providing details on the implementation and parallelization of code.

Daisuke Ishii, Kazuki Yoshizoe, Toyotaro Suzumura

We present a scalable parallel solver for numerical constraint satisfaction problems (NCSPs). Our parallelization scheme consists of homogeneous worker solvers, each of which runs on an available core and communicates with others via the global load balancing (GLB) method. The parallel solver is implemented with X10 that provides an implementation of GLB as a library. In experiments, several NCSPs from the literature were solved and attained up to 516-fold speedup using 600 cores of the TSUBAME2.5 supercomputer.

Daisuke Ishii, Kazuki Yoshizoe, Toyotaro Suzumura

We present a parallel solver for numerical constraint satisfaction problems (NCSPs) that can scale on a number of cores. Our proposed method runs worker solvers on the available cores and simultaneously the workers cooperate for the search space distribution and balancing. In the experiments, we attained up to 119-fold speedup using 256 cores of a parallel computer.