arXiv subjects

Yun Liu

Publications and source records attributed to Yun Liu.

At least 235 records · Page 13Linked to original sources

arXiv2021

Coarse-grained decomposition and fine-grained interaction for multi-hop question answering

Recent advances regarding question answering and reading comprehension have resulted in models that surpass human performance when the answer is contained in a single, continuous passage of text, requiring only single-hop reasoning. However, in actual scenarios, lots of complex queries require multi-hop reasoning. The key to the Question Answering task is semantic feature interaction between documents and questions, which is widely processed by Bi-directional Attention Flow (Bi-DAF), but Bi-DAF generally captures only the surface semantics of words in complex questions and fails to capture implied semantic feature of intermediate answers. As a result, Bi-DAF partially ignores part of the contexts related to the question and cannot extract the most important parts of multiple documents. In this paper we propose a new model architecture for multi-hop question answering, by applying two completion strategies: (1) Coarse-Grain complex question Decomposition (CGDe) strategy are introduced to decompose complex question into simple ones under the condition of without any additional annotations (2) Fine-Grained Interaction (FGIn) strategy are introduced to better represent each word in the document and extract more comprehensive and accurate sentences related to the inference path. The above two strategies are combined and tested on the SQuAD and HotpotQA datasets, and the experimental results show that our method outperforms state-of-the-art baselines.

cs.CL↗

arXiv2021

Optical and electronic properties of colloidal CdSe Quantum Rings

Luminescent colloidal CdSe nanorings are a new type of semiconductor structure that have attracted interest due to the potential for unique physics arising from their non-trivial toroidal shape. However, the exciton properties and dynamics of these materials with complex topology are not yet well understood. Here, we use a combination of femtosecond vibrational spectroscopy, temperature-resolved photoluminescence (PL), and single particle measurements to study these materials. We find that on transformation of CdSe nanoplatelets to nanorings, by perforating the center of platelets, the emission lifetime decreases and the emission spectrum broadens due to ensemble variations in the ring size and thickness. The reduced PL quantum yield of nanorings (~10%) compared to platelets (~30%) is attributed to an enhanced coupling between: (i) excitons and CdSe LO-phonons at 200 cm-1 and (ii) negatively charged selenium-rich traps which give nanorings a high surface charge (~-50 mV). Population of these weakly emissive trap sites dominates the emission properties with an increased trap emission at low temperatures relative to excitonic emission. Our results provide a detailed picture of the nature of excitons in nanorings and the influence of phonons and surface charge in explaining the broad shape of the PL spectrum and the origin of PL quantum yield losses. Furthermore, they suggest that the excitonic properties of nanorings are not solely a consequence of the toroidal shape but are also a result of traps introduced by puncturing the platelet center.

physics.app-ph↗

arXiv2020

Predicting Prostate Cancer-Specific Mortality with A.I.-based Gleason Grading

Gleason grading of prostate cancer is an important prognostic factor but suffers from poor reproducibility, particularly among non-subspecialist pathologists. Although artificial intelligence (A.I.) tools have demonstrated Gleason grading on-par with expert pathologists, it remains an open question whether A.I. grading translates to better prognostication. In this study, we developed a system to predict prostate-cancer specific mortality via A.I.-based Gleason grading and subsequently evaluated its ability to risk-stratify patients on an independent retrospective cohort of 2,807 prostatectomy cases from a single European center with 5-25 years of follow-up (median: 13, interquartile range 9-17). The A.I.'s risk scores produced a C-index of 0.84 (95%CI 0.80-0.87) for prostate cancer-specific mortality. Upon discretizing these risk scores into risk groups analogous to pathologist Grade Groups (GG), the A.I. had a C-index of 0.82 (95%CI 0.78-0.85). On the subset of cases with a GG in the original pathology report (n=1,517), the A.I.'s C-indices were 0.87 and 0.85 for continuous and discrete grading, respectively, compared to 0.79 (95%CI 0.71-0.86) for GG obtained from the reports. These represent improvements of 0.08 (95%CI 0.01-0.15) and 0.07 (95%CI 0.00-0.14) respectively. Our results suggest that A.I.-based Gleason grading can lead to effective risk-stratification and warrants further evaluation for improving disease management.

cs.CV↗

arXiv2020

Detecting hidden signs of diabetes in external eye photographs

Diabetes-related retinal conditions can be detected by examining the posterior of the eye. By contrast, examining the anterior of the eye can reveal conditions affecting the front of the eye, such as changes to the eyelids, cornea, or crystalline lens. In this work, we studied whether external photographs of the front of the eye can reveal insights into both diabetic retinal diseases and blood glucose control. We developed a deep learning system (DLS) using external eye photographs of 145,832 patients with diabetes from 301 diabetic retinopathy (DR) screening sites in one US state, and evaluated the DLS on three validation sets containing images from 198 sites in 18 other US states. In validation set A (n=27,415 patients, all undilated), the DLS detected poor blood glucose control (HbA1c > 9%) with an area under receiver operating characteristic curve (AUC) of 70.2; moderate-or-worse DR with an AUC of 75.3; diabetic macular edema with an AUC of 78.0; and vision-threatening DR with an AUC of 79.4. For all 4 prediction tasks, the DLS's AUC was higher (p<0.001) than using available self-reported baseline characteristics (age, sex, race/ethnicity, years with diabetes). In terms of positive predictive value, the predicted top 5% of patients had a 67% chance of having HbA1c > 9%, and a 20% chance of having vision threatening diabetic retinopathy. The results generalized to dilated pupils (validation set B, 5,058 patients) and to a different screening service (validation set C, 10,402 patients). Our results indicate that external eye photographs contain information useful for healthcare providers managing patients with diabetes, and may help prioritize patients for in-person screening. Further work is needed to validate these findings on different devices and patient populations (those without diabetes) to evaluate its utility for remote diagnosis and management.

eess.IV↗

arXiv2020

Interpretable Survival Prediction for Colorectal Cancer using Deep Learning

Deriving interpretable prognostic features from deep-learning-based prognostic histopathology models remains a challenge. In this study, we developed a deep learning system (DLS) for predicting disease specific survival for stage II and III colorectal cancer using 3,652 cases (27,300 slides). When evaluated on two validation datasets containing 1,239 cases (9,340 slides) and 738 cases (7,140 slides) respectively, the DLS achieved a 5-year disease-specific survival AUC of 0.70 (95%CI 0.66-0.73) and 0.69 (95%CI 0.64-0.72), and added significant predictive value to a set of 9 clinicopathologic features. To interpret the DLS, we explored the ability of different human-interpretable features to explain the variance in DLS scores. We observed that clinicopathologic features such as T-category, N-category, and grade explained a small fraction of the variance in DLS scores (R2=18% in both validation sets). Next, we generated human-interpretable histologic features by clustering embeddings from a deep-learning based image-similarity model and showed that they explain the majority of the variance (R2 of 73% to 80%). Furthermore, the clustering-derived feature most strongly associated with high DLS scores was also highly prognostic in isolation. With a distinct visual appearance (poorly differentiated tumor cell clusters adjacent to adipose tissue), this feature was identified by annotators with 87.0-95.5% accuracy. Our approach can be used to explain predictions from a prognostic deep learning model and uncover potentially-novel prognostic features that can be reliably identified by people for future validation studies.

eess.IV↗

arXiv2020

Sub-Band Knowledge Distillation Framework for Speech Enhancement

In single-channel speech enhancement, methods based on full-band spectral features have been widely studied. However, only a few methods pay attention to non-full-band spectral features. In this paper, we explore a knowledge distillation framework based on sub-band spectral mapping for single-channel speech enhancement. Specifically, we divide the full frequency band into multiple sub-bands and pre-train an elite-level sub-band enhancement model (teacher model) for each sub-band. These teacher models are dedicated to processing their own sub-bands. Next, under the teacher models' guidance, we train a general sub-band enhancement model (student model) that works for all sub-bands. Without increasing the number of model parameters and computational complexity, the student model's performance is further improved. To evaluate our proposed method, we conducted a large number of experiments on an open-source data set. The final experimental results show that the guidance from the elite-level teacher models dramatically improves the student model's performance, which exceeds the full-band model by employing fewer parameters.

eess.AS↗

arXiv2020

Efficacious symmetry-adapted atomic displacement method for lattice dynamical studies

Small displacement methods have been successfully used to calculate the lattice dynamical properties of crystals. It involves displacing atoms by a small amount in order to calculate the induced forces on all atoms in a supercell for the computation of force constants. Even though these methods are widely in use, to our knowledge, there is no systematic discussion of optimal displacement directions from the crystal's symmetry point of view nor a rigorous error analysis of such methods. Based on the group theory and point group symmetry of a crystal, we propose displacement directions, with an equivalent concept of the group of $k$, deduced directly in the Cartesian coordinates rather than the usual fractional coordinates, that maintain the theoretical maximum for the triple product $V$ spanned by the three displacements to avoid possible severe roundoff errors. The proposed displacement directions are generated from a minimal set of irreducible atomic displacements that keep the required independent force calculations to a minimum. We find the error in the calculated force constant explicitly depends on the inverse of $V$ and inaccuracy of the forces. Test systems such as Si, graphene, and orthorhombic Sb2S3 are used to illustrate the method. Our displacement method is shown to be very robust in treating low-symmetry cells with a large `aspect ratio' due to huge differences in lattice parameters, use of a large vacuum height, or a very oblique unit cell due to unconventional choice of primitive lattice vectors. It is expected that our displacement strategy can be used to address higher-order interatomic interactions to achieve good accuracy and efficiency.

cond-mat.mtrl-sci↗

arXiv2020

DCCRN: Deep Complex Convolution Recurrent Network for Phase-Aware Speech Enhancement

Speech enhancement has benefited from the success of deep learning in terms of intelligibility and perceptual quality. Conventional time-frequency (TF) domain methods focus on predicting TF-masks or speech spectrum, via a naive convolution neural network (CNN) or recurrent neural network (RNN). Some recent studies use complex-valued spectrogram as a training target but train in a real-valued network, predicting the magnitude and phase component or real and imaginary part, respectively. Particularly, convolution recurrent network (CRN) integrates a convolutional encoder-decoder (CED) structure and long short-term memory (LSTM), which has been proven to be helpful for complex targets. In order to train the complex target more effectively, in this paper, we design a new network structure simulating the complex-valued operation, called Deep Complex Convolution Recurrent Network (DCCRN), where both CNN and RNN structures can handle complex-valued operation. The proposed DCCRN models are very competitive over other previous networks, either on objective or subjective metric. With only 3.7M parameters, our DCCRN models submitted to the Interspeech 2020 Deep Noise Suppression (DNS) challenge ranked first for the real-time-track and second for the non-real-time track in terms of Mean Opinion Score (MOS).

eess.AS↗

arXiv2020

MS-TCN++: Multi-Stage Temporal Convolutional Network for Action Segmentation

With the success of deep learning in classifying short trimmed videos, more attention has been focused on temporally segmenting and classifying activities in long untrimmed videos. State-of-the-art approaches for action segmentation utilize several layers of temporal convolution and temporal pooling. Despite the capabilities of these approaches in capturing temporal dependencies, their predictions suffer from over-segmentation errors. In this paper, we propose a multi-stage architecture for the temporal action segmentation task that overcomes the limitations of the previous approaches. The first stage generates an initial prediction that is refined by the next ones. In each stage we stack several layers of dilated temporal convolutions covering a large receptive field with few parameters. While this architecture already performs well, lower layers still suffer from a small receptive field. To address this limitation, we propose a dual dilated layer that combines both large and small receptive fields. We further decouple the design of the first stage from the refining stages to address the different requirements of these stages. Extensive evaluation shows the effectiveness of the proposed model in capturing long-range dependencies and recognizing action segments. Our models achieve state-of-the-art results on three datasets: 50Salads, Georgia Tech Egocentric Activities (GTEA), and the Breakfast dataset.

cs.CV↗

arXiv2020

Generalized Zero-Shot Learning via VAE-Conditioned Generative Flow

Generalized zero-shot learning (GZSL) aims to recognize both seen and unseen classes by transferring knowledge from semantic descriptions to visual representations. Recent generative methods formulate GZSL as a missing data problem, which mainly adopts GANs or VAEs to generate visual features for unseen classes. However, GANs often suffer from instability, and VAEs can only optimize the lower bound on the log-likelihood of observed data. To overcome the above limitations, we resort to generative flows, a family of generative models with the advantage of accurate likelihood estimation. More specifically, we propose a conditional version of generative flows for GZSL, i.e., VAE-Conditioned Generative Flow (VAE-cFlow). By using VAE, the semantic descriptions are firstly encoded into tractable latent distributions, conditioned on that the generative flow optimizes the exact log-likelihood of the observed visual features. We ensure the conditional latent distribution to be both semantic meaningful and inter-class discriminative by i) adopting the VAE reconstruction objective, ii) releasing the zero-mean constraint in VAE posterior regularization, and iii) adding a classification regularization on the latent variables. Our method achieves state-of-the-art GZSL results on five well-known benchmark datasets, especially for the significant improvement in the large-scale setting. Code is released at https://github.com/guyuchao/VAE-cFlow-ZSL.

cs.CV↗

arXiv2020

Predicting Risk of Developing Diabetic Retinopathy using Deep Learning

Diabetic retinopathy (DR) screening is instrumental in preventing blindness, but faces a scaling challenge as the number of diabetic patients rises. Risk stratification for the development of DR may help optimize screening intervals to reduce costs while improving vision-related outcomes. We created and validated two versions of a deep learning system (DLS) to predict the development of mild-or-worse ("Mild+") DR in diabetic patients undergoing DR screening. The two versions used either three-fields or a single field of color fundus photographs (CFPs) as input. The training set was derived from 575,431 eyes, of which 28,899 had known 2-year outcome, and the remaining were used to augment the training process via multi-task learning. Validation was performed on both an internal validation set (set A; 7,976 eyes; 3,678 with known outcome) and an external validation set (set B; 4,762 eyes; 2,345 with known outcome). For predicting 2-year development of DR, the 3-field DLS had an area under the receiver operating characteristic curve (AUC) of 0.79 (95%CI, 0.78-0.81) on validation set A. On validation set B (which contained only a single field), the 1-field DLS's AUC was 0.70 (95%CI, 0.67-0.74). The DLS was prognostic even after adjusting for available risk factors (p<0.001). When added to the risk factors, the 3-field DLS improved the AUC from 0.72 (95%CI, 0.68-0.76) to 0.81 (95%CI, 0.77-0.84) in validation set A, and the 1-field DLS improved the AUC from 0.62 (95%CI, 0.58-0.66) to 0.71 (95%CI, 0.68-0.75) in validation set B. The DLSs in this study identified prognostic information for DR development from CFPs. This information is independent of and more informative than the available risk factors.

eess.IV↗

arXiv2020

Rethinking RGB-D Salient Object Detection: Models, Data Sets, and Large-Scale Benchmarks

The use of RGB-D information for salient object detection has been extensively explored in recent years. However, relatively few efforts have been put towards modeling salient object detection in real-world human activity scenes with RGBD. In this work, we fill the gap by making the following contributions to RGB-D salient object detection. (1) We carefully collect a new SIP (salient person) dataset, which consists of ~1K high-resolution images that cover diverse real-world scenes from various viewpoints, poses, occlusions, illuminations, and backgrounds. (2) We conduct a large-scale (and, so far, the most comprehensive) benchmark comparing contemporary methods, which has long been missing in the field and can serve as a baseline for future research. We systematically summarize 32 popular models and evaluate 18 parts of 32 models on seven datasets containing a total of about 97K images. (3) We propose a simple general architecture, called Deep Depth-Depurator Network (D3Net). It consists of a depth depurator unit (DDU) and a three-stream feature learning module (FLM), which performs low-quality depth map filtering and cross-modal feature learning respectively. These components form a nested structure and are elaborately designed to be learned jointly. D3Net exceeds the performance of any prior contenders across all five metrics under consideration, thus serving as a strong model to advance research in this field. We also demonstrate that D3Net can be used to efficiently extract salient object masks from real scenes, enabling effective background changing application with a speed of 65fps on a single GPU. All the saliency maps, our new SIP dataset, the D3Net model, and the evaluation tools are publicly available at https://github.com/DengPingFan/D3NetBenchmark.

cs.CV↗

arXiv2020

Deep Learning for Strong Lensing Search: Tests of the Convolutional Neural Networks and New Candidates from KiDS DR3

Convolutional Neutral Networks have been successfully applied in searching for strong lensing systems, leading to discoveries of new candidates from large surveys. On the other hand, systematic investigations about their robustness are still lacking. In this paper, we first construct a neutral network, and apply it to $r$-band images of Luminous Red Galaxies (LRGs) of the Kilo Degree Survey (KiDS) Data Release 3 to search for strong lensing systems. We build two sets of training samples, one fully from simulations, and the other one using the LRG stamps from KiDS observations as the foreground lens images. With the former training sample, we find 48 high probability candidates after human-inspection, and among them, 27 are newly identified. Using the latter training set, about 67\% of the aforementioned 48 candidates are also found, and there are 11 more new strong lensing candidates identified. We then carry out tests on the robustness of the network performance with respect to the variation of PSF. With the testing samples constructed using PSF in the range of 0.4 to 2 times of the median PSF of the training sample, we find that our network performs rather stable, and the degradation is small. We also investigate how the volume of the training set can affect our network performance by varying it from 0.1 millions to 0.8 millions. The output results are rather stable showing that within the considered range, our network performance is not very sensitive to the volume size.

astro-ph.CO↗

arXiv2020

Magnetic ordering and spin dynamics in $S=5/2$ staggered triangular lattice antiferromagnet Ba$_2$MnTeO$_6$

We report studies of the magnetic properties of a staggered stacked triangular lattice Ba$_2$MnTeO$_6$ using magnetic susceptibility, specific heat, neutron powder diffraction and inelastic neutron scattering measurements, as well as first principles density functional theory calculations. Neutron diffraction measurements reveal an antiferromagnetic order with a propagated vector $\textbf{\emph{k}}=(0.5, 0.5, 0)$ and N{é}el transition temperature of $T_\text{N}\approx20$ K. The dominant interaction derived from the Curie-Weiss fitting to the inverse DC susceptibility is antiferromagnetic. Through modelling the INS spectrum with the linear spin wave theory, the magnetic exchange interactions for the nearest intralayer, nearest interlayer, and next nearest interlayer are determined to be $J_1=0.27(3),J_2=0.27(3),$ and $J_3=-0.05(1)$ meV, respectively, and a small value of easy-axis anisotropy of $D_{zz}=-0.01$ meV is introduced. We derive a magnetic phase diagram that reveals that it is the competition between $J_1, J_2$, and $J_3$ that stabilizes the collinear stripe-type antiferromagnetic order.

cond-mat.str-el↗

arXiv2020

Laser-assisted high-energy proton pulse extraction for feasibility study of co-located muon source at the SNS

We have experimentally demonstrated the first non-intrusive 1-GeV proton beam extraction for the generation of muons with a temporal structure optimized for Muon Spin Relaxation/Rotation/Resonance (MuSR) applications. The proton pulses are extracted based on the laser neutralization of 1 GeV hydrogen ion (H-) beam in the high energy beam transport of the Spallation Neutron Source (SNS) accelerator. The maximum flux of the extracted proton beam accounts for only 0.2% of the total proton beam used for neutron production, a marked difference from the 20% reduction at other co-located muon and neutron facilities, and thus the proposed method will result in negligible impact on the SNS operation. This paper describes the development of a fiber/solid-state hybrid laser system that has high flexibility of pulse structure and output power, initial experiments on laser neutralization of H- beam and separation of H0 beam from the existing SNS accelerator beam line, conversion of H0 to proton at the SNS linac dump, and measurement results of 30 ns/50 kHz proton pulses. This system conclusively demonstrates the feasibility of laser-based proton beam extraction to power a world-leading MuSR facility at the SNS.

physics.acc-ph↗

arXiv2020

Tunable Optoelectronic Properties of WS$_2$ by Local Strain Engineering and Folding

Local strain engineering is an exciting approach to tune the optoelectronic properties of materials. Two dimensional (2D) materials such as 2D transition metal dichalcogenides (TMDs) are particularly well suited for this purpose because they have high flexibility and they can withstand high deformations before rupture. Local strain engineering in 2D TMDs is achieved via strained wrinkles. Wrinkles on thick layers of TMDs are reported to show interesting photoluminescence enhancement due to bandgap modulation and funneling effect. However, the wrinkles in ultrathin TMDs have not been investigated because they can easily fall down to form folds in these ultrathin layers of TMDCs. Here, we have achieved both wrinkles and folds simultaneously in 1-3L WS2 using a new fabrication technique. A layer dependent reduction in surface potential is found for both folded layers and perfect pack layers due to the dominant interlayer screening effect. Strain dependent modulation in semi conductive junction properties is observed for strain induced wrinkles through current scanning. Thermo-ionic modelling suggests that the strained (1.6%) wrinkles can lower the Schottky barrier height (SBH) by 20%. Upon illumination, SBH reduces significantly due to photo-generated carriers. Our results present an important advance towards controlling the optoelectronic properties of atomically thin WS2 via strain engineering, with applications in optoelectronics, quantum optics and nanophotonics device fabrication.

physics.optics↗

arXiv2019

Ordered or Orderless: A Revisit for Video based Person Re-Identification

Is recurrent network really necessary for learning a good visual representation for video based person re-identification (VPRe-id)? In this paper, we first show that the common practice of employing recurrent neural networks (RNNs) to aggregate temporal spatial features may not be optimal. Specifically, with a diagnostic analysis, we show that the recurrent structure may not be effective to learn temporal dependencies than what we expected and implicitly yields an orderless representation. Based on this observation, we then present a simple yet surprisingly powerful approach for VPRe-id, where we treat VPRe-id as an efficient orderless ensemble of image based person re-identification problem. More specifically, we divide videos into individual images and re-identify person with ensemble of image based rankers. Under the i.i.d. assumption, we provide an error bound that sheds light upon how could we improve VPRe-id. Our work also presents a promising way to bridge the gap between video and image based person re-identification. Comprehensive experimental evaluations demonstrate that the proposed solution achieves state-of-the-art performances on multiple widely used datasets (iLIDS-VID, PRID 2011, and MARS).

cs.CV↗

arXiv2019

Deep learning-based survival prediction for multiple cancer types using histopathology images

Prognostic information at diagnosis has important implications for cancer treatment and monitoring. Although cancer staging, histopathological assessment, molecular features, and clinical variables can provide useful prognostic insights, improving risk stratification remains an active research area. We developed a deep learning system (DLS) to predict disease specific survival across 10 cancer types from The Cancer Genome Atlas (TCGA). We used a weakly-supervised approach without pixel-level annotations, and tested three different survival loss functions. The DLS was developed using 9,086 slides from 3,664 cases and evaluated using 3,009 slides from 1,216 cases. In multivariable Cox regression analysis of the combined cohort including all 10 cancers, the DLS was significantly associated with disease specific survival (hazard ratio of 1.58, 95% CI 1.28-1.70, p<0.0001) after adjusting for cancer type, stage, age, and sex. In a per-cancer adjusted subanalysis, the DLS remained a significant predictor of survival in 5 of 10 cancer types. Compared to a baseline model including stage, age, and sex, the c-index of the model demonstrated an absolute 3.7% improvement (95% CI 1.0-6.5) in the combined cohort. Additionally, our models stratified patients within individual cancer stages, particularly stage II (p=0.025) and stage III (p<0.001). By developing and evaluating prognostic models across multiple cancer types, this work represents one of the most comprehensive studies exploring the direct prediction of clinical outcomes using deep learning and histopathology images. Our analysis demonstrates the potential for this approach to provide prognostic information in multiple cancer types, and even within specific pathologic stages. However, given the relatively small number of clinical events, we observed wide confidence intervals, suggesting that future work will benefit from larger datasets.

q-bio.QM↗