Search

Inspired by Large Language Models (LLMs), Time Series Forecasting (TSF), a long-standing task in time series analysis, is undergoing a transition towards Large Time Series Models (LTSMs), aiming to train universal transformer-based models for TSF. However, training LTSMs on heterogeneous time series data poses unique challenges, including diverse frequencies, dimensions, and patterns across datasets. Recent endeavors have studied and evaluated various design choices aimed at enhancing LTSM training and generalization capabilities, spanning pre-processing techniques, model configurations, and dataset configurations. In this work, we comprehensively analyze these design choices and aim to identify the best practices for training LTSM. Moreover, we propose \emph{time series prompt}, a novel statistical prompting strategy tailored to time series data. Furthermore, based on the observations in our analysis, we introduce \texttt{LTSM-bundle}, which bundles the best design choices we have identified. Empirical results demonstrate that \texttt{LTSM-bundle} achieves superior zero-shot and few-shot performances compared to state-of-the-art LSTMs and traditional TSF methods on benchmark datasets.

The theory of two-sided matching has been extensively developed and applied to many real-life application domains. As the theory has been applied to increasingly diverse types of environments, researchers and practitioners have encountered various forms of distributional constraints. As a mechanism can handle a more general class of constraints, we can assign students more flexibly to colleges to increase students' welfare. However, it turns out that there exists a trade-off between students' welfare (efficiency) and fairness (which means no student has justified envy). Furthermore, this trade-off becomes sharper as the class of constraints becomes more general. The first contribution of this paper is to clarify the boundary on whether a strategyproof and fair mechanism can satisfy certain efficiency properties for each class of constraints. Our second contribution is to establish a weaker fairness requirement called envy-freeness up to $k$ peers (EF-$k$), which is inspired by a similar concept used in the fair division of indivisible items. EF-$k$ guarantees that each student has justified envy towards at most $k$ students. By varying $k$, EF-$k$ can represent different levels of fairness. We investigate theoretical properties associated with EF-$k$. Furthermore, we develop two contrasting strategyproof mechanisms that work for general hereditary constraints, i.e., one mechanism can guarantee a strong efficiency requirement, while the other can guarantee EF-$k$ for any fixed $k$. We evaluate the performance of these mechanisms through computer simulation., Comment: 11 pages

In real-world applications, machine learning models often become obsolete due to shifts in the joint distribution arising from underlying temporal trends, a phenomenon known as the "concept drift". Existing works propose model-specific strategies to achieve temporal generalization in the near-future domain. However, the diverse characteristics of real-world datasets necessitate customized prediction model architectures. To this end, there is an urgent demand for a model-agnostic temporal domain generalization approach that maintains generality across diverse data modalities and architectures. In this work, we aim to address the concept drift problem from a data-centric perspective to bypass considering the interaction between data and model. Developing such a framework presents non-trivial challenges: (i) existing generative models struggle to generate out-of-distribution future data, and (ii) precisely capturing the temporal trends of joint distribution along chronological source domains is computationally infeasible. To tackle the challenges, we propose the COncept Drift simulAtor (CODA) framework incorporating a predicted feature correlation matrix to simulate future data for model training. Specifically, CODA leverages feature correlations to represent data characteristics at specific time points, thereby circumventing the daunting computational costs. Experimental results demonstrate that using CODA-generated data as training input effectively achieves temporal domain generalization across different model architectures.

The theory of two-sided matching has been extensively developed and applied to many real-life application domains. As the theory has been applied to increasingly diverse types of environments, researchers and practitioners have encountered various forms of distributional constraints. Arguably, the most general class of distributional constraints would be hereditary constraints; if a matching is feasible, then any matching that assigns weakly fewer students at each college is also feasible. However, under general hereditary constraints, it is shown that no strategyproof mechanism exists that simultaneously satisfies fairness and weak nonwastefulness, which is an efficiency (students' welfare) requirement weaker than nonwastefulness. We propose a new strategyproof mechanism that works for hereditary constraints called the Multi-Stage Generalized Deferred Acceptance mechanism (MS-GDA). It uses the Generalized Deferred Acceptance mechanism (GDA) as a subroutine, which works when distributional constraints belong to a well-behaved class called hereditary M$^\natural$-convex set. We show that GDA satisfies several desirable properties, most of which are also preserved in MS-GDA. We experimentally show that MS-GDA strikes a good balance between fairness and efficiency (students' welfare) compared to existing strategyproof mechanisms when distributional constraints are close to an M$^\natural$-convex set., Comment: 23 pages

The exponential growth in scholarly publications necessitates advanced tools for efficient article retrieval, especially in interdisciplinary fields where diverse terminologies are used to describe similar research. Traditional keyword-based search engines often fall short in assisting users who may not be familiar with specific terminologies. To address this, we present a knowledge graph-based paper search engine for biomedical research to enhance the user experience in discovering relevant queries and articles. The system, dubbed DiscoverPath, employs Named Entity Recognition (NER) and part-of-speech (POS) tagging to extract terminologies and relationships from article abstracts to create a KG. To reduce information overload, DiscoverPath presents users with a focused subgraph containing the queried entity and its neighboring nodes and incorporates a query recommendation system, enabling users to iteratively refine their queries. The system is equipped with an accessible Graphical User Interface that provides an intuitive visualization of the KG, query recommendations, and detailed article information, enabling efficient article retrieval, thus fostering interdisciplinary knowledge exploration. DiscoverPath is open-sourced at https://github.com/ynchuang/DiscoverPath.

Graph Neural Networks (GNNs) have achieved state-of-the-art performance in recommender systems. Nevertheless, the process of searching and ranking from a large item corpus usually requires high latency, which limits the widespread deployment of GNNs in industry-scale applications. To address this issue, many methods compress user/item representations into the binary embedding space to reduce space requirements and accelerate inference. Also, they use the Straight-through Estimator (STE) to prevent vanishing gradients during back-propagation. However, the STE often causes the gradient mismatch problem, leading to sub-optimal results. In this work, we present the Hessian-aware Quantized GNN (HQ-GNN) as an effective solution for discrete representations of users/items that enable fast retrieval. HQ-GNN is composed of two components: a GNN encoder for learning continuous node embeddings and a quantized module for compressing full-precision embeddings into low-bit ones. Consequently, HQ-GNN benefits from both lower memory requirements and faster inference speeds compared to vanilla GNNs. To address the gradient mismatch problem in STE, we further consider the quantized errors and its second-order derivatives for better stability. The experimental results on several large-scale datasets show that HQ-GNN achieves a good balance between latency and performance.

Imbalanced datasets are commonly observed in various real-world applications, presenting significant challenges in training classifiers. When working with large datasets, the imbalanced issue can be further exacerbated, making it exceptionally difficult to train classifiers effectively. To address the problem, over-sampling techniques have been developed to linearly interpolating data instances between minorities and their neighbors. However, in many real-world scenarios such as anomaly detection, minority instances are often dispersed diversely in the feature space rather than clustered together. Inspired by domain-agnostic data mix-up, we propose generating synthetic samples iteratively by mixing data samples from both minority and majority classes. It is non-trivial to develop such a framework, the challenges include source sample selection, mix-up strategy selection, and the coordination between the underlying model and mix-up strategies. To tackle these challenges, we formulate the problem of iterative data mix-up as a Markov decision process (MDP) that maps data attributes onto an augmentation strategy. To solve the MDP, we employ an actor-critic framework to adapt the discrete-continuous decision space. This framework is utilized to train a data augmentation policy and design a reward signal that explores classifier uncertainty and encourages performance improvement, irrespective of the classifier's convergence. We demonstrate the effectiveness of our proposed framework through extensive experiments conducted on seven publicly available benchmark datasets using three different types of classifiers. The results of these experiments showcase the potential and promise of our framework in addressing imbalanced datasets with diverse minorities.

Despite the impressive prediction ability, machine learning models show discrimination towards certain demographics and suffer from unfair prediction behaviors. To alleviate the discrimination, extensive studies focus on eliminating the unequal distribution of sensitive attributes via multiple approaches. However, due to privacy concerns, sensitive attributes are often either unavailable or missing in real-world scenarios. Therefore, several existing works alleviate the bias without sensitive attributes. Those studies face challenges, either in inaccurate predictions of sensitive attributes or the need to mitigate unequal distribution of manually defined non-sensitive attributes related to bias. The latter requires strong assumptions about the correlation between sensitive and non-sensitive attributes. As data distribution and task goals vary, the strong assumption on non-sensitive attributes may not be valid and require domain expertise. In this work, we propose an assumption-free framework to detect the related attributes automatically by modeling feature interaction for bias mitigation. The proposed framework aims to mitigate the unfair impact of identified biased feature interactions. Experimental results on four real-world datasets demonstrate that our proposed framework can significantly alleviate unfair prediction behaviors by considering biased feature interactions.

In a usual dark matter (DM) model without a huge mass difference between the DM and lighter mediator, using the coupling strength suitable for having the correct relic density, the resulting self-interaction becomes several orders of magnitude smaller than that required to interpret the small scale structures. We present a framework that can offer a solution for this point. We consider a model that contains the vector DM and a heavier but unstable Higgs-like scalar in the hidden sector. When the temperature drops below $\sim m_{\rm DM}$, the hidden sector, which is thermally decoupled from the visible sector, enters a cannibal phase, during which the DM density is depleted with the out-of-equilibrium decay of the scalar. The favored parameter region, giving the correct relic density and the proper size of self-interactions, shows the scalar-to-DM mass ratio $\in [1.1,1.33]$ and the scalar mass $\in[9,114]\,{\rm MeV}$. A sizable parameter space still survives the most current constraints and can be further probed by the near future NA62 beam dump experiment., Comment: 28 pages, 8 figures, The content revised; Some statements corrected; Main results unchanged

Time series anomaly detection is a challenging task with a wide range of real-world applications. Due to label sparsity, training a deep anomaly detector often relies on unsupervised approaches. Recent efforts have been devoted to time series domain adaptation to leverage knowledge from similar domains. However, existing solutions may suffer from negative knowledge transfer on anomalies due to their diversity and sparsity. Motivated by the empirical study of context alignment between two domains, we aim to transfer knowledge between two domains via adaptively sampling context information for two domains. This is challenging because it requires simultaneously modeling the complex in-domain temporal dependencies and cross-domain correlations while exploiting label information from the source domain. To this end, we propose a framework that combines context sampling and anomaly detection into a joint learning procedure. We formulate context sampling into the Markov decision process and exploit deep reinforcement learning to optimize the time series domain adaptation process via context sampling and design a tailored reward function to generate domain-invariant features that better align two domains for anomaly detection. Experiments on three public datasets show promise for knowledge transfer between two similar domains and two entirely different domains.

Artificial Intelligence (AI) is making a profound impact in almost every domain. A vital enabler of its great success is the availability of abundant and high-quality data for building machine learning models. Recently, the role of data in AI has been significantly magnified, giving rise to the emerging concept of data-centric AI. The attention of researchers and practitioners has gradually shifted from advancing model design to enhancing the quality and quantity of the data. In this survey, we discuss the necessity of data-centric AI, followed by a holistic view of three general data-centric goals (training data development, inference data development, and data maintenance) and the representative methods. We also organize the existing literature from automation and collaboration perspectives, discuss the challenges, and tabulate the benchmarks for various tasks. We believe this is the first comprehensive survey that provides a global view of a spectrum of tasks across various stages of the data lifecycle. We hope it can help the readers efficiently grasp a broad picture of this field, and equip them with the techniques and further research ideas to systematically engineer data for building AI systems. A companion list of data-centric AI resources will be regularly updated on https://github.com/daochenzha/data-centric-AI, Comment: 38 pages, 6 figues, 5 tables. A companion list of data-centric AI resources is available at https://github.com/daochenzha/data-centric-AI

The role of data in building AI systems has recently been significantly magnified by the emerging concept of data-centric AI (DCAI), which advocates a fundamental shift from model advancements to ensuring data quality and reliability. Although our community has continuously invested efforts into enhancing data in different aspects, they are often isolated initiatives on specific tasks. To facilitate the collective initiative in our community and push forward DCAI, we draw a big picture and bring together three general missions: training data development, inference data development, and data maintenance. We provide a top-level discussion on representative DCAI tasks and share perspectives. Finally, we list open challenges. More resources are summarized at https://github.com/daochenzha/data-centric-AI, Comment: Accepted by SDM 2023 Blue Sky Track. More resources are summarized at https://github.com/daochenzha/data-centric-AI

Knowledge graph data are prevalent in real-world applications, and knowledge graph neural networks (KGNNs) are essential techniques for knowledge graph representation learning. Although KGNN effectively models the structural information from knowledge graphs, these frameworks amplify the underlying data bias that leads to discrimination towards certain groups or individuals in resulting applications. Additionally, as existing debiasing approaches mainly focus on the entity-wise bias, eliminating the multi-hop relational bias that pervasively exists in knowledge graphs remains an open question. However, it is very challenging to eliminate relational bias due to the sparsity of the paths that generate the bias and the non-linear proximity structure of knowledge graphs. To tackle the challenges, we propose Fair-KGNN, a KGNN framework that simultaneously alleviates multi-hop bias and preserves the proximity information of entity-to-relation in knowledge graphs. The proposed framework is generalizable to mitigate the relational bias for all types of KGNN. We develop two instances of Fair-KGNN incorporating with two state-of-the-art KGNN models, RGCN and CompGCN, to mitigate gender-occupation and nationality-salary bias. The experiments carried out on three benchmark knowledge graph datasets demonstrate that the Fair-KGNN can effectively mitigate unfair situations during representation learning while preserving the predictive performance of KGNN models.

We study embedding table placement for distributed recommender systems, which aims to partition and place the tables on multiple hardware devices (e.g., GPUs) to balance the computation and communication costs. Although prior work has explored learning-based approaches for the device placement of computational graphs, embedding table placement remains to be a challenging problem because of 1) the operation fusion of embedding tables, and 2) the generalizability requirement on unseen placement tasks with different numbers of tables and/or devices. To this end, we present DreamShard, a reinforcement learning (RL) approach for embedding table placement. DreamShard achieves the reasoning of operation fusion and generalizability with 1) a cost network to directly predict the costs of the fused operation, and 2) a policy network that is efficiently trained on an estimated Markov decision process (MDP) without real GPU execution, where the states and the rewards are estimated with the cost network. Equipped with sum and max representation reductions, the two networks can directly generalize to any unseen tasks with different numbers of tables and/or devices without fine-tuning. Extensive experiments show that DreamShard substantially outperforms the existing human expert and RNN-based strategies with up to 19% speedup over the strongest baseline on large-scale synthetic tables and our production tables. The code is available at https://github.com/daochenzha/dreamshard, Comment: Accepted by NeurIPS 2022

Kinematically forbidden channels can set the freeze-out dark matter (DM) relic abundance. These channels are described by DM annihilations into heavier states, which vanish at zero temperature limit, but occur at finite temperatures in the early Universe. For the case that the final state of the forbidden channel is scalar mediators that couple to Standard Model (SM) matter through mixing with the SM Higgs, the signals from DM-nucleon interactions and from mediator-related missing energy or displaced vertices could be detected by direct detections and particle physics experiments, respectively. We thus present a study on the simplest secluded vector dark matter model that can exhibit this scenario in the mass range from sub-GeV to TeV. The dark matter resides in the hidden sector, which is in thermal equilibrium with the SM before freeze-out. During freeze-out, the depletion of its density results from its annihilation into two heavier but metastable scalars, where the coupling can be determined by having the correct relic density and constrained by the perturbative unitarity bound. However, much of the allowed parameter space is insensitive to the mixing angle between the hidden scalar and SM Higgs. We find that a more significant mass splitting between DM and the mediator can be allowed only in the sub-GeV region. Moreover, the mass splitting in the TeV region is required to be within the percent level. This model of the forbidden DM interacting with SM particles through the scalar portal is testable in experiments., Comment: 35 pages, 8 figures, 1 table; Freeze-in curves in Fig. 1 revised; Table 1 updated; Fig. 7 (new) added; Fig. 8 updated; discussions about the large mass splitting corrected

Imbalanced learning is a fundamental challenge in data mining, where there is a disproportionate ratio of training samples in each class. Over-sampling is an effective technique to tackle imbalanced learning through generating synthetic samples for the minority class. While numerous over-sampling algorithms have been proposed, they heavily rely on heuristics, which could be sub-optimal since we may need different sampling strategies for different datasets and base classifiers, and they cannot directly optimize the performance metric. Motivated by this, we investigate developing a learning-based over-sampling algorithm to optimize the classification performance, which is a challenging task because of the huge and hierarchical decision space. At the high level, we need to decide how many synthetic samples to generate. At the low level, we need to determine where the synthetic samples should be located, which depends on the high-level decision since the optimal locations of the samples may differ for different numbers of samples. To address the challenges, we propose AutoSMOTE, an automated over-sampling algorithm that can jointly optimize different levels of decisions. Motivated by the success of SMOTE~\cite{chawla2002smote} and its extensions, we formulate the generation process as a Markov decision process (MDP) consisting of three levels of policies to generate synthetic samples within the SMOTE search space. Then we leverage deep hierarchical reinforcement learning to optimize the performance metric on the validation data. Extensive experiments on six real-world datasets demonstrate that AutoSMOTE significantly outperforms the state-of-the-art resampling algorithms. The code is at https://github.com/daochenzha/autosmote, Comment: Accepted by CIKM 2022

We study time-series classification (TSC), a fundamental task of time-series data mining. Prior work has approached TSC from two major directions: (1) similarity-based methods that classify time-series based on the nearest neighbors, and (2) deep learning models that directly learn the representations for classification in a data-driven manner. Motivated by the different working mechanisms within these two research lines, we aim to connect them in such a way as to jointly model time-series similarities and learn the representations. This is a challenging task because it is unclear how we should efficiently leverage similarity information. To tackle the challenge, we propose Similarity-Aware Time-Series Classification (SimTSC), a conceptually simple and general framework that models similarity information with graph neural networks (GNNs). Specifically, we formulate TSC as a node classification problem in graphs, where the nodes correspond to time-series, and the links correspond to pair-wise similarities. We further design a graph construction strategy and a batch training algorithm with negative sampling to improve training efficiency. We instantiate SimTSC with ResNet as the backbone and Dynamic Time Warping (DTW) as the similarity measure. Extensive experiments on the full UCR datasets and several multivariate datasets demonstrate the effectiveness of incorporating similarity information into deep learning models in both supervised and semi-supervised settings. Our code is available at https://github.com/daochenzha/SimTSC, Comment: Accepted by SDM 2022

Action recognition is an important task for video understanding with broad applications. However, developing an effective action recognition solution often requires extensive engineering efforts in building and testing different combinations of the modules and their hyperparameters. In this demo, we present AutoVideo, a Python system for automated video action recognition. AutoVideo is featured for 1) highly modular and extendable infrastructure following the standard pipeline language, 2) an exhaustive list of primitives for pipeline construction, 3) data-driven tuners to save the efforts of pipeline tuning, and 4) easy-to-use Graphical User Interface (GUI). AutoVideo is released under MIT license at https://github.com/datamllab/autovideo, Comment: Accepted by IJCAI https://github.com/datamllab/autovideo

Supervised regression to demonstrations has been demonstrated to be a stable way to train deep policy networks. We are motivated to study how we can take full advantage of supervised loss functions for stably training deep reinforcement learning agents. This is a challenging task because it is unclear how the training data could be collected to enable policy improvement. In this work, we propose Self-Supervised Reinforcement Learning (SSRL), a simple algorithm that optimizes policies with purely supervised losses. We demonstrate that, without policy gradient or value estimation, an iterative procedure of ``labeling" data and supervised regression is sufficient to drive stable policy improvement. By selecting and imitating trajectories with high episodic rewards, SSRL is surprisingly competitive to contemporary algorithms with more stable performance and less running time, showing the potential of solving reinforcement learning with supervised learning techniques. The code is available at https://github.com/daochenzha/SSRL

The public GALPROP code gives a fully numerical solution for the spectrum of cosmic rays propagating through a linear Galactic wind directed outwards from the Galactic plane in the framework of a cylindrical diffusion model. Alternatively, for the linear Galactic wind case, we present a semi-analytical solution of the two-zone cylindrical model which describes the spectrum of cosmic-ray antiprotons produced from the primary sources throughout the dark matter halo. While the secondary antiprotons can be generated by, e.g., the GALPROP WebRun, consistently using this semi-analytical solution for the primary ones would be helpful to build a suited dark matter (DM) model, which may be sophisticated, and then to perform a statistical analysis when confronting with data. As an example, using the obtained formula, we study the possible DM signal, through the channel $\text{DM DM} \to {\bar b}b$, and its constraint from the measurement of the AMS-02 antiproton-to-proton ratio. The indication of the DM signal is discussed. The advantage of the semi-analytical approach and its comparison with the GALPROP model are presented., Comment: 18 pages, 1 Table, 7 Figures. Accepted for publication in ApJ

We present TODS, an automated Time Series Outlier Detection System for research and industrial applications. TODS is a highly modular system that supports easy pipeline construction. The basic building block of TODS is primitive, which is an implementation of a function with hyperparameters. TODS currently supports 70 primitives, including data processing, time series processing, feature analysis, detection algorithms, and a reinforcement module. Users can freely construct a pipeline using these primitives and perform end- to-end outlier detection with the constructed pipeline. TODS provides a Graphical User Interface (GUI), where users can flexibly design a pipeline with drag-and-drop. Moreover, a data-driven searcher is provided to automatically discover the most suitable pipelines given a dataset. TODS is released under Apache 2.0 license at https://github.com/datamllab/tods., Comment: Accepted by AAAI'21 demo track

High false-positive rate is a long-standing challenge for anomaly detection algorithms, especially in high-stake applications. To identify the true anomalies, in practice, analysts or domain experts will be employed to investigate the top instances one by one in a ranked list of anomalies identified by an anomaly detection system. This verification procedure generates informative labels that can be leveraged to re-rank the anomalies so as to help the analyst to discover more true anomalies given a time budget. Some re-ranking strategies have been proposed to approximate the above sequential decision process. Specifically, existing strategies have been focused on making the top instances more likely to be anomalous based on the feedback. Then they greedily select the top-1 instance for query. However, these greedy strategies could be sub-optimal since some low-ranked instances could be more helpful in the long-term. In this work, we propose Active Anomaly Detection with Meta-Policy (Meta-AAD), a novel framework that learns a meta-policy for query selection. Specifically, Meta-AAD leverages deep reinforcement learning to train the meta-policy to select the most proper instance to explicitly optimize the number of discovered anomalies throughout the querying process. Meta-AAD is easy to deploy since a trained meta-policy can be directly applied to any new datasets without further tuning. Extensive experiments on 24 benchmark datasets demonstrate that Meta-AAD significantly outperforms the state-of-the-art re-ranking strategies and the unsupervised baseline. The empirical analysis shows that the trained meta-policy is transferable and inherently achieves a balance between long-term and short-term rewards., Comment: Accepted by ICDM 2020

Graph data are pervasive in many real-world applications. Recently, increasing attention has been paid on graph neural networks (GNNs), which aim to model the local graph structures and capture the hierarchical patterns by aggregating the information from neighbors with stackable network modules. Motivated by the observation that different nodes often require different iterations of aggregation to fully capture the structural information, in this paper, we propose to explicitly sample diverse iterations of aggregation for different nodes to boost the performance of GNNs. It is a challenging task to develop an effective aggregation strategy for each node, given complex graphs and sparse features. Moreover, it is not straightforward to derive an efficient algorithm since we need to feed the sampled nodes into different number of network layers. To address the above challenges, we propose Policy-GNN, a meta-policy framework that models the sampling procedure and message passing of GNNs into a combined learning process. Specifically, Policy-GNN uses a meta-policy to adaptively determine the number of aggregations for each node. The meta-policy is trained with deep reinforcement learning (RL) by exploiting the feedback from the model. We further introduce parameter sharing and a buffer mechanism to boost the training efficiency. Experimental results on three real-world benchmark datasets suggest that Policy-GNN significantly outperforms the state-of-the-art alternatives, showing the promise in aggregation optimization for GNNs., Comment: Accepted by ACM SIGKDD'20 research track

Policy distillation, which transfers a teacher policy to a student policy has achieved great success in challenging tasks of deep reinforcement learning. This teacher-student framework requires a well-trained teacher model which is computationally expensive. Moreover, the performance of the student model could be limited by the teacher model if the teacher model is not optimal. In the light of collaborative learning, we study the feasibility of involving joint intellectual efforts from diverse perspectives of student models. In this work, we introduce dual policy distillation(DPD), a student-student framework in which two learners operate on the same environment to explore different perspectives of the environment and extract knowledge from each other to enhance their learning. The key challenge in developing this dual learning framework is to identify the beneficial knowledge from the peer learner for contemporary learning-based reinforcement learning algorithms, since it is unclear whether the knowledge distilled from an imperfect and noisy peer learner would be helpful. To address the challenge, we theoretically justify that distilling knowledge from a peer learner will lead to policy improvement and propose a disadvantageous distillation strategy based on the theoretical results. The conducted experiments on several continuous control tasks show that the proposed framework achieves superior performance with a learning-based agent and function approximation without the use of expensive teacher models., Comment: Accepeted by IJCAI'20

We show the presence of a potentially detectable gamma-ray line in the Fermi Galactic center excess in light of the secluded (vector) dark matter (DM) model in which the hidden scalar, nearly degenerate with DM in mass, mediates the interaction of the secluded DM with the Standard Model (SM) due to its mixing with the SM Higgs. We find that the parameter region $m_X\in[60, 132]$ GeV can provide a good fit to the Fermi Galactic center gamma-ray excess spectrum, appearing a prominent gamma-ray line with the energy $\in [30, 66]$ GeV. The best fit gives $m_X\simeq m_S \simeq 86$ GeV with a $p$-value$\, =0.42$, so that the resultant gamma-ray line, arising from the decay of the scalar mediator into $\gamma\gamma$, peaks at 43 GeV. We derive constraints on the annihilation cross section from the Fermi-LAT gamma-ray line search, gamma-ray observations of the Fermi-LAT dwarf spheroidal galaxies, and Planck cosmic microwave background measurement. For the secluded vector DM model, the parameter space constrained by the current XENON1T and future LUX-ZEPLIN is shown. Finally, for the mixing angle between the Higgs sectors, we discuss its lower bound, which is required by the big bang nucleosynthesis constraint and relevant to the hidden sector decoupling temperature., Comment: 39, pages, 11 figures, 1 table; typos corrected; minor changes to match published version in JHEP

RLCard is an open-source toolkit for reinforcement learning research in card games. It supports various card environments with easy-to-use interfaces, including Blackjack, Leduc Hold'em, Texas Hold'em, UNO, Dou Dizhu and Mahjong. The goal of RLCard is to bridge reinforcement learning and imperfect information games, and push forward the research of reinforcement learning in domains with multiple agents, large state and action space, and sparse reward. In this paper, we provide an overview of the key components in RLCard, a discussion of the design principles, a brief introduction of the interfaces, and comprehensive evaluations of the environments. The codes and documents are available at https://github.com/datamllab/rlcard, Comment: AAAI-20 Workshop on Reinforcement Learning in Games

Experience replay enables reinforcement learning agents to memorize and reuse past experiences, just as humans replay memories for the situation at hand. Contemporary off-policy algorithms either replay past experiences uniformly or utilize a rule-based replay strategy, which may be sub-optimal. In this work, we consider learning a replay policy to optimize the cumulative reward. Replay learning is challenging because the replay memory is noisy and large, and the cumulative reward is unstable. To address these issues, we propose a novel experience replay optimization (ERO) framework which alternately updates two policies: the agent policy, and the replay policy. The agent is updated to maximize the cumulative reward based on the replayed data, while the replay policy is updated to provide the agent with the most useful experiences. The conducted experiments on various continuous control tasks demonstrate the effectiveness of ERO, empirically showing promise in experience replay learning to improve the performance of off-policy reinforcement learning algorithms.

The secluded dark matter resides within a hidden sector and self-annihilates into lighter mediators which subsequently decay to the Standard Model (SM) particles. Depending on the coupling strength of the mediator to the SM, the hidden sector can be kinetically decoupled from the SM bath when the temperature drops below the mediator's mass, and the dark matter annihilation cross section at freeze-out is thus possible to be boosted above the conventional value of weak interacting massive particles. We present a comprehensive study on thermodynamic evolution of the hidden sector from the first principle, using the simplest secluded vector dark matter model. Motivated by the observation of Galactic center gamma-ray excess, we take two mass sets $\sim{\cal O}(80\, \text{GeV})$ for the dark matter and mediator as examples to illustrate the thermodynamics. The coupled Boltzmann moment equations for number densities and temperature evolutions of the hidden sector are numerically solved. The formalism can be easily extended to a general secluded dark matter model. We show that a long-lived mediator can result in a boosted dark matter annihilation cross section to account for the relic abundance. We further show the parameter space which provides a good fit to the Galactic center excess data and is compatible with the current bounds and LUX-ZEPLIN projected sensitivity. We find that the future observations of dwarf spheroidal galaxies offer promising reach to probe the most relic allowed parameter space relevant to the boosted dark matter annihilation cross section., Comment: 51 pages, 15 figures; some typos and equations corrected

Storing and processing massive small files is one of the major challenges for the Hadoop Distributed File System (HDFS). In order to provide fast data access, the NameNode (NN) in HDFS maintains the metadata of all files in its main-memory. Hadoop performs well with a small number of large files that require relatively little metadata in the NN s memory. But for a large number of small files, Hadoop has problems such as NN memory overload caused by the huge metadata size of these small files. We present a new type of archive file, Hadoop Perfect File (HPF), to solve HDFS s small files problem by merging small files into a large file on HDFS. Existing archive files offer limited functionality and have poor performance when accessing a file in the merged file due to the fact that during metadata lookup it is necessary to read and process the entire index file(s). In contrast, HPF file can directly access the metadata of a particular file from its index file without having to process it entirely. The HPF index system uses two hash functions: file s metadata are distributed through index files by using a dynamic hash function and, for each index file, we build an order preserving perfect hash function that preserves the position of each file s metadata in the index file. The HPF design will only read the part of the index file that contains the metadata of the searched file during its access. HPF file also supports the file appending functionality after its creation. Our experiments show that HPF can be more than 40% faster file s access from the original HDFS. If we don t consider the caching effect, HPF s file access is around 179% faster than MapFile and 11294% faster than HAR file. If we consider caching effect, HPF is around 35% faster than MapFile and 105% faster than HAR file.

Cross-domain collaborative filtering (CF) aims to alleviate data sparsity in single-domain CF by leveraging knowledge transferred from related domains. Many traditional methods focus on enriching compared neighborhood relations in CF directly to address the sparsity problem. In this paper, we propose superhighway construction, an alternative explicit relation-enrichment procedure, to improve recommendations by enhancing cross-domain connectivity. Specifically, assuming partially overlapped items (users), superhighway bypasses multi-hop inter-domain paths between cross-domain users (items, respectively) with direct paths to enrich the cross-domain connectivity. The experiments conducted on a real-world cross-region music dataset and a cross-platform movie dataset show that the proposed superhighway construction significantly improves recommendation performance in both target and source domains.

Image perception is one of the most direct ways to provide contextual information about a user concerning his/her surrounding environment; hence images are a suitable proxy for contextual recommendation. We propose a novel representation learning framework for image-based music recommendation that bridges the heterogeneity gap between music and image data; the proposed method is a key component for various contextual recommendation tasks. Preliminary experiments show that for an image-to-song retrieval task, the proposed method retrieves relevant or conceptually similar songs for input images., Comment: 2 pages, LBRS@RecSys'18

We have built a lepton-specific next-to-minimal two-Higgs-doublet-portal vector dark matter model. The vector dark matter in the hidden sector does not directly couple to the visible sector, but instead annihilates into the hidden Higgs bosons which decay through a small coupling into the CP-odd Higgs bosons. In this model, the Galactic center gamma-ray excess is mainly due to the 2-step cascade annihilation with $\tau$'s in the final state. The obtained mass of the CP-odd Higgs $A$ in the Galactic center excess fit can explain the muon $g-2$ anomaly at the 2$\sigma$ level without violating the stringent constraints from the lepton universality and $\tau$ decays. We show three different freeze-out types of the dark matter relic, called (i) the conventional WIMP dark matter, (ii) the unconventional WIMP dark matter and (iii) the cannibally co-decaying dark matter, depending on the magnitudes of the mixing angles between the hidden Higgs and visible two-Higgs doublets. The dark matter in the hidden sector is secluded from detections in the direct searches or colliders, while the dark matter annihilation signals are not suppressed in a general hidden sector dark matter model. We discuss the constraints from observations of the dwarf spheroidal galaxies and the Fermi-LAT projected sensitivity., Comment: 47 pages, 1 table, 10 figures, minor revision. Matches published version

In light of the observed Galactic center gamma-ray excess, we investigate a simplified model, for which the scalar dark matter interacts with quarks through a pseudoscalar mediator. The viable regions of the parameter space, that can also account for the relic density and evade the current searches, are identified, if the low-velocity dark matter annihilates through an $s$-channel off-shell mediator mostly into $\bar{b} b$, and/or annihilates directly into two ${\it hidden}$ on-shell mediators, which subsequently decay into the quark pairs. These two kinds of annihilations are $s$ wave. The projected monojet limit set by the high luminosity LHC sensitivity could constrain the favored parameter space, where the mediator's mass is larger than the dark matter mass by a factor of 2. We show that the projected sensitivity of 15-year Fermi-LAT observations of dwarf spheroidal galaxies can provide a stringent constraint on the most parameter space allowed in this model. If the on-shell mediator channel contributes to the dark matter annihilation cross sections over 50$\%$, this model with a lighter mediator can be probed in the projected PICO-500L experiment., Comment: 31 pages, 1 table, 5 figures, typo in Eq. (7) corrected, caption of Fig. 4 revised

We present a study for two-quark light-cone distribution amplitudes for the $1^3D_2$ light tensor meson states with quantum number $J^{PC}=2^{--}$. Because of the G-parity, the chiral-even two-quark light-cone distribution amplitudes of this tensor meson are antisymmetric under the interchange of momentum fractions of the quark and antiquark in the SU(3) limit, while the chiral-odd ones are symmetric. The asymptotic leading-twist LCDAs with the strange quark mass correction are shown. We estimate the relevant parameters, the decay constants $f_T$ and $f_T^\perp$, and first Gegenbauer moment $a_1^\perp$, by using the QCD sum rule method. These parameters play a central role in the investigation of $B$ meson decaying into the $2^{--}$ tensor mesons., Comment: 18 pages, 3 Figures