arXiv subjects

Qian Li

Publications and source records attributed to Qian Li.

At least 271 records · Page 15Linked to original sources

arXiv2021

Question-Aware Memory Network for Multi-hop Question Answering in Human-Robot Interaction

Knowledge graph question answering is an important technology in intelligent human-robot interaction, which aims at automatically giving answer to human natural language question with the given knowledge graph. For the multi-relation question with higher variety and complexity, the tokens of the question have different priority for the triples selection in the reasoning steps. Most existing models take the question as a whole and ignore the priority information in it. To solve this problem, we propose question-aware memory network for multi-hop question answering, named QA2MN, to update the attention on question timely in the reasoning process. In addition, we incorporate graph context information into knowledge graph embedding model to increase the ability to represent entities and relations. We use it to initialize the QA2MN model and fine-tune it in the training process. We evaluate QA2MN on PathQuestion and WorldCup2014, two representative datasets for complex multi-hop question answering. The result demonstrates that QA2MN achieves state-of-the-art Hits@1 accuracy on the two datasets, which validates the effectiveness of our model.

cs.CL↗

arXiv2021

Few-shot Continual Learning: a Brain-inspired Approach

It is an important yet challenging setting to continually learn new tasks from a few examples. Although numerous efforts have been devoted to either continual learning or few-shot learning, little work has considered this new setting of few-shot continual learning (FSCL), which needs to minimize the catastrophic forgetting to the old tasks and gradually improve the ability of few-shot generalization. In this paper, we provide a first systematic study on FSCL and present an effective solution with deep neural networks. Our solution is based on the observation that continual learning of a task sequence inevitably interferes few-shot generalization, which makes it highly nontrivial to extend few-shot learning strategies to continual learning scenarios. We draw inspirations from the robust brain system and develop a method that (1) interdependently updates a pair of fast / slow weights for continual learning and few-shot learning to disentangle their divergent objectives, inspired by the biological model of meta-plasticity and fast / slow synapse; and (2) applies a brain-inspired two-step consolidation strategy to learn a task sequence without forgetting in the fast weights while improve generalization without overfitting in the slow weights. Extensive results on various benchmarks show that our method achieves a better performance than joint training of all the tasks ever seen. The ability of few-shot generalization is also substantially improved from incoming tasks and examples.

cs.LG↗

arXiv2021

Ionic liquid gating induced two superconductor-insulator phase transitions in spinel oxide Li$_{1 \pm x}$Ti$_2$O$_{4-δ}$

The associations between emergent physical phenomena (e.g., superconductivity) and orbital, charge, and spin degrees of freedom of $3d$ electrons are intriguing in transition metal compounds. Here, we successfully manipulate the superconductivity of spinel oxide Li$_{1\pm x}$Ti$_2$O$_{4-δ}$ (LTO) by ionic liquid gating. A dome-shaped superconducting phase diagram is established, where two insulating phases are disclosed both in heavily electron-doping and hole-doping regions. The superconductor-insulator transition (SIT) in the hole-doping region can be attributed to the loss of Ti valence electrons. In the electron-doping region, LTO exhibits an unexpected SIT instead of a metallic behavior despite an increase in carrier density. Furthermore, a thermal hysteresis is observed in the normal state resistance curve, suggesting a first-order phase transition. We speculate that the SIT and the thermal hysteresis stem from the enhanced $3d$ electron correlations and the formation of orbital ordering by comparing the transport and structural results of LTO with the other spinel oxide superconductor MgTi$_2$O$_4$, as well as analysing the electronic structure by first-principles calculations. Further comprehension of the detailed interplay between superconductivity and orbital ordering would contribute to the revealing of unconventional superconducting pairing mechanism.

cond-mat.supr-con↗

arXiv2021

A General Framework for Learning Prosodic-Enhanced Representation of Rap Lyrics

Learning and analyzing rap lyrics is a significant basis for many web applications, such as music recommendation, automatic music categorization, and music information retrieval, due to the abundant source of digital music in the World Wide Web. Although numerous studies have explored the topic, knowledge in this field is far from satisfactory, because critical issues, such as prosodic information and its effective representation, as well as appropriate integration of various features, are usually ignored. In this paper, we propose a hierarchical attention variational autoencoder framework (HAVAE), which simultaneously consider semantic and prosodic features for rap lyrics representation learning. Specifically, the representation of the prosodic features is encoded by phonetic transcriptions with a novel and effective strategy~(i.e., rhyme2vec). Moreover, a feature aggregation strategy is proposed to appropriately integrate various features and generate prosodic-enhanced representation. A comprehensive empirical evaluation demonstrates that the proposed framework outperforms the state-of-the-art approaches under various metrics in different rap lyrics learning tasks.

cs.CL↗

arXiv2021

Subterahertz collective dynamics of polar vortices

The collective dynamics of topological structures have been of great interest from both fundamental and applied perspectives. For example, the studies of dynamical properties of magnetic vortices and skyrmions not only deepened the understanding of many-body physics but also led to potential applications in data processing and storage. Topological structures constructed from electrical polarization rather than spin have recently been realized in ferroelectric superlattices, promising for ultrafast electric-field control of topological orders. However, little is known about the dynamics of such complex extended nanostructures which in turn underlies their functionalities. Using terahertz-field excitation and femtosecond x-ray diffraction measurements, we observe ultrafast collective polarization dynamics that are unique to polar vortices, with orders of magnitude higher frequencies and smaller lateral size than those of experimentally realized magnetic vortices. A previously unseen soft mode, hereafter referred to as a vortexon, emerges as transient arrays of nanoscale circular patterns of atomic displacements, which reverse their vorticity on picosecond time scales. Its frequency is significantly reduced at a critical strain, indicating a condensation of structural dynamics. First-principles-based atomistic calculations and phase-field modeling reveal the microscopic atomic arrangements and frequencies of the vortex modes. The discovery of subterahertz collective dynamics in polar vortices opens up opportunities for applications of electric-field driven data processing in topological structures with ultrahigh speed and density.

cond-mat.mes-hall↗

arXiv2021

Magnetic field-tuned quantum criticality in optimally electron-doped cuprate thin films

Antiferromagnetic (AF) spin fluctuations are commonly believed to play a key role in electron pairing of cuprate superconductors. In electron-doped cuprates, it is still in paradox about the interplay among different electronic states in quantum perturbations, especially between superconducting and magnetic states. Here, we report a systematic transport study on cation-optimized La2-xCexCuO4 (x = 0.10) thin films in high magnetic fields. We find an AF quantum phase transition near 60 T, where the Hall number jumps from nH =-x to nH = 1-x, resembling the change of nH at the AF boundary (xAF = 0.14) tuned by Ce doping. In the AF region a spin dependent state manifesting anomalous positive magnetoresistance is observed, which is closely related to superconductivity. Once the AF state is suppressed by magnetic field, a polarized ferromagnetic state is predicted, reminiscent of the recently reported ferromagnetic state at the quantum endpoint of the superconducting dome by Ce doping. The magnetic field that drives phase transitions in a similar but distinct manner to doping thereby provides a unique perspective to understand the quantum criticality of electron-doped cuprates.

cond-mat.supr-con↗

arXiv2020

INFaaS: A Model-less and Managed Inference Serving System

Despite existing work in machine learning inference serving, ease-of-use and cost efficiency remain challenges at large scales. Developers must manually search through thousands of model-variants -- versions of already-trained models that differ in hardware, resource footprints, latencies, costs, and accuracies -- to meet the diverse application requirements. Since requirements, query load, and applications themselves evolve over time, these decisions need to be made dynamically for each inference query to avoid excessive costs through naive autoscaling. To avoid navigating through the large and complex trade-off space of model-variants, developers often fix a variant across queries, and replicate it when load increases. However, given the diversity across variants and hardware platforms in the cloud, a lack of understanding of the trade-off space can incur significant costs to developers. This paper introduces INFaaS, a managed and model-less system for distributed inference serving, where developers simply specify the performance and accuracy requirements for their applications without needing to specify a specific model-variant for each query. INFaaS generates model-variants, and efficiently navigates the large trade-off space of model-variants on behalf of developers to meet application-specific objectives: (a) for each query, it selects a model, hardware architecture, and model optimizations, (b) it combines VM-level horizontal autoscaling with model-level autoscaling, where multiple, different model-variants are used to serve queries within each machine. By leveraging diverse variants and sharing hardware resources across models, INFaaS achieves 1.3x higher throughput, violates latency objectives 1.6x less often, and saves up to 21.6x in cost (8.5x on average) compared to state-of-the-art inference serving systems on AWS EC2.

cs.DC↗

arXiv2020

Consistency of a kind of general noncanonical warm inflation

The framework of a kind of noncanonical warm inflation is introduced, and the dynamical equations of this scenario are presented. We propose the slow roll approximations and give some redefining slow roll parameters in this scenario which remain dimensionless. Performing systemic stability analysis, we calculate the slow roll conditions to guarantee that slow roll approximations hold. The slow roll conditions suggest slow roll inflation in general noncanonical warm inflationary scenario can still exist, and in addition, the slow roll approximations are more easily to be satisfied. Then, a concrete Dirac-Born-Infeld warm inflationary model is studied.

gr-qc↗

arXiv2020

Pixel-Semantic Revise of Position Learning A One-Stage Object Detector with A Shared Encoder-Decoder

Recently, many methods have been proposed for object detection. They cannot detect objects by semantic features, adaptively. In this work, according to channel and spatial attention mechanisms, we mainly analyze that different methods detect objects adaptively. Some state-of-the-art detectors combine different feature pyramids with many mechanisms to enhance multi-level semantic information. However, they require more cost. This work addresses that by an anchor-free detector with shared encoder-decoder with attention mechanism, extracting shared features. We consider features of different levels from backbone (e.g., ResNet-50) as the basis features. Then, we feed the features into a simple module, followed by a detector header to detect objects. Meantime, we use the semantic features to revise geometric locations, and the detector is a pixel-semantic revising of position. More importantly, this work analyzes the impact of different pooling strategies (e.g., mean, maximum or minimum) on multi-scale objects, and finds the minimum pooling improve detection performance on small objects better. Compared with state-of-the-art MNC based on ResNet-101 for the standard MSCOCO 2014 baseline, our method improves detection AP of 3.8%.

cs.CV↗

arXiv2020

Video Face Recognition System: RetinaFace-mnet-faster and Secondary Search

Face recognition is widely used in the scene. However, different visual environments require different methods, and face recognition has a difficulty in complex environments. Therefore, this paper mainly experiments complex faces in the video. First, we design an image pre-processing module for fuzzy scene or under-exposed faces to enhance images. Our experimental results demonstrate that effective images pre-processing improves the accuracy of 0.11%, 0.2% and 1.4% on LFW, WIDER FACE and our datasets, respectively. Second, we propose RetinacFace-mnet-faster for detection and a confidence threshold specification for face recognition, reducing the lost rate. Our experimental results show that our RetinaFace-mnet-faster for 640*480 resolution on the Tesla P40 and single-thread improve speed of 16.7% and 70.2%, respectively. Finally, we design secondary search mechanism with HNSW to improve performance. Ours is suitable for large-scale datasets, and experimental results show that our method is 82% faster than the violent retrieval for the single-frame detection.

cs.CV↗

arXiv2020

Leveraging Multi-level Dependency of Relational Sequences for Social Spammer Detection

Much recent research has shed light on the development of the relation-dependent but content-independent framework for social spammer detection. This is largely because the relation among users is difficult to be altered when spammers attempt to conceal their malicious intents. Our study investigates the spammer detection problem in the context of multi-relation social networks, and makes an attempt to fully exploit the sequences of heterogeneous relations for enhancing the detection accuracy. Specifically, we present the Multi-level Dependency Model (MDM). The MDM is able to exploit user's long-term dependency hidden in their relational sequences along with short-term dependency. Moreover, MDM fully considers short-term relational sequences from the perspectives of individual-level and union-level, due to the fact that the type of short-term sequences is multi-folds. Experimental results on a real-world multi-relational social network demonstrate the effectiveness of our proposed MDM on multi-relational social spammer detection.

cs.LG↗

arXiv2020

How to Investigate the Historical Roots and Evolution of Research Fields in China? A Case Study on iMetrics Using RootCite

This paper aimed to provide an approach to investigate the historical roots and evolution of research fields in China by extending the reference publication year spectroscopy (RPYS). RootCite, an open source software accepts raw data from both the Web of Science and the China Social Science Citation Index (CSSCI), was developed using python. We took iMetrics in China as the research case. 5,141 Chinese iMetrics related publications with 73,376 non-distinct cited references (CR) collected from the CSSCI were analyzed using RootCite. The results showed that the first CR in the field can be dated back to 1882 and written in English; but the majority (64.2%) of the CR in the field were Chinese publications. 17 peaks referring to 18 seminal works (13 in English and 5 in Chinese) were located during the period from 1900 to 2017. The field shared the same roots with that in the English world but has its own characteristics, and it was then shaped by contributions from both the English world and China. The five Chinese works have played irreplaceable and positive roles in the historical evolutionary path of the field, which should not be ignored, especially for the evolution of the field. This research demonstrated how RootCite aided the task of identifying the origin and evolution of research fields in China, which could be valuable for extending RPYS for countries with other languages.

cs.DL↗

arXiv2020

Region-Referenced Spectral Power Dynamics of EEG Signals: A Hierarchical Modeling Approach

Functional brain imaging through electroencephalography (EEG) relies upon the analysis and interpretation of high-dimensional, spatially organized time series. We propose to represent time-localized frequency domain characterizations of EEG data as region-referenced functional data. This representation is coupled with a hierarchical modeling approach to multivariate functional observations. Within this familiar setting, we discuss how several prior models relate to structural assumptions about multivariate covariance operators. An overarching modeling framework, based on infinite factorial decompositions, is finally proposed to balance flexibility and efficiency in estimation. The motivating application stems from a study of implicit auditory learning, in which typically developing (TD) children, and children with autism spectrum disorder (ASD) were exposed to a continuous speech stream. Using the proposed model, we examine differential band power dynamics as brain function is interrogated throughout the duration of a computer-controlled experiment. Our work offers a novel look at previous findings in psychiatry, and provides further insights into the understanding of ASD. Our approach to inference is fully Bayesian and implemented in a highly optimized Rcpp package.

stat.AP↗

arXiv2020

Learning Differential Diagnosis of Skin Conditions with Co-occurrence Supervision using Graph Convolutional Networks

Skin conditions are reported the 4th leading cause of nonfatal disease burden worldwide. However, given the colossal spectrum of skin disorders defined clinically and shortage in dermatology expertise, diagnosing skin conditions in a timely and accurate manner remains a challenging task. Using computer vision technologies, a deep learning system has proven effective assisting clinicians in image diagnostics of radiology, ophthalmology and more. In this paper, we propose a deep learning system (DLS) that may predict differential diagnosis of skin conditions using clinical images. Our DLS formulates the differential diagnostics as a multi-label classification task over 80 conditions when only incomplete image labels are available. We tackle the label incompleteness problem by combining a classification network with a Graph Convolutional Network (GCN) that characterizes label co-occurrence and effectively regularizes it towards a sparse representation. Our approach is demonstrated on 136,462 clinical images and concludes that the classification accuracy greatly benefit from the Co-occurrence supervision. Our DLS achieves 93.6% top-5 accuracy on 12,378 test images and consistently outperform the baseline classification network.

cs.CV↗

arXiv2020

Stochastic Batch Augmentation with An Effective Distilled Dynamic Soft Label Regularizer

Data augmentation have been intensively used in training deep neural network to improve the generalization, whether in original space (e.g., image space) or representation space. Although being successful, the connection between the synthesized data and the original data is largely ignored in training, without considering the distribution information that the synthesized samples are surrounding the original sample in training. Hence, the behavior of the network is not optimized for this. However, that behavior is crucially important for generalization, even in the adversarial setting, for the safety of the deep learning system. In this work, we propose a framework called Stochastic Batch Augmentation (SBA) to address these problems. SBA stochastically decides whether to augment at iterations controlled by the batch scheduler and in which a ''distilled'' dynamic soft label regularization is introduced by incorporating the similarity in the vicinity distribution respect to raw samples. The proposed regularization provides direct supervision by the KL-Divergence between the output soft-max distributions of original and virtual data. Our experiments on CIFAR-10, CIFAR-100, and ImageNet show that SBA can improve the generalization of the neural networks and speed up the convergence of network training.

cs.LG↗

arXiv2020

Triaging moderate COVID-19 and other viral pneumonias from routine blood tests

The COVID-19 is sweeping the world with deadly consequences. Its contagious nature and clinical similarity to other pneumonias make separating subjects contracted with COVID-19 and non-COVID-19 viral pneumonia a priority and a challenge. However, COVID-19 testing has been greatly limited by the availability and cost of existing methods, even in developed countries like the US. Intrigued by the wide availability of routine blood tests, we propose to leverage them for COVID-19 testing using the power of machine learning. Two proven-robust machine learning model families, random forests (RFs) and support vector machines (SVMs), are employed to tackle the challenge. Trained on blood data from 208 moderate COVID-19 subjects and 86 subjects with non-COVID-19 moderate viral pneumonia, the best result is obtained in an SVM-based classifier with an accuracy of 84%, a sensitivity of 88%, a specificity of 80%, and a precision of 92%. The results are found explainable from both machine learning and medical perspectives. A privacy-protected web portal is set up to help medical personnel in their practice and the trained models are released for developers to further build other applications. We hope our results can help the world fight this pandemic and welcome clinical verification of our approach on larger populations.

cs.LG↗

arXiv2020

U-Net Using Stacked Dilated Convolutions for Medical Image Segmentation

This paper proposes a novel U-Net variant using stacked dilated convolutions for medical image segmentation (SDU-Net). SDU-Net adopts the architecture of vanilla U-Net with modifications in the encoder and decoder operations (an operation indicates all the processing for feature maps of the same resolution). Unlike vanilla U-Net which incorporates two standard convolutions in each encoder/decoder operation, SDU-Net uses one standard convolution followed by multiple dilated convolutions and concatenates all dilated convolution outputs as input to the next operation. Experiments showed that SDU-Net outperformed vanilla U-Net, attention U-Net (AttU-Net), and recurrent residual U-Net (R2U-Net) in all four tested segmentation tasks while using parameters around 40% of vanilla U-Net's, 17% of AttU-Net's, and 15% of R2U-Net's.

eess.IV↗

arXiv2020

Weakly Supervised Context Encoder using DICOM metadata in Ultrasound Imaging

Modern deep learning algorithms geared towards clinical adaption rely on a significant amount of high fidelity labeled data. Low-resource settings pose challenges like acquiring high fidelity data and becomes the bottleneck for developing artificial intelligence applications. Ultrasound images, stored in Digital Imaging and Communication in Medicine (DICOM) format, have additional metadata data corresponding to ultrasound image parameters and medical exams. In this work, we leverage DICOM metadata from ultrasound images to help learn representations of the ultrasound image. We demonstrate that the proposed method outperforms the non-metadata based approaches across different downstream tasks.

eess.IV↗