arXiv subjects
Publications and source records attributed to Xin Zhang.
In this note, we provide a smooth variational principle on Wasserstein space by constructing a smooth gauge-type function using the sliced Wasserstein distance. This function is a crucial tool for optimization problems and in viscosity theory of PDEs on Wasserstein space.
Background estimation is essential when studying TeV gamma-ray astronomy for extensive air shower arrays. In this work, by applying four applying four different methods including equi-zenith angle method, surrounding window method, direct integration method, and time-swapping method, the number of the background events is calculated. Based on simulation samples, the statistical significance of the excess signal from different background estimation methods is determined. Following this, we discuss the limits and the applicability of the four methods under different conditions. Under the detector stability assumption with signal, the results from the above four methods are consistent at the 1 sigma level. In the no signal condition, when the acceptance of the detector changes with both space and time, the surrounding window method is most stable and hardly affected. In this acceptance assumption, we find that the background estimation in the direct integration and time-swapping methods are sensitive to the selection of time window, and the shorter time window can reduce the impact on the background estimation to some extent.
Stuttering is a speech disorder where the natural flow of speech is interrupted by blocks, repetitions or prolongations of syllables, words and phrases. The majority of existing automatic speech recognition (ASR) interfaces perform poorly on utterances with stutter, mainly due to lack of matched training data. Synthesis of speech with stutter thus presents an opportunity to improve ASR for this type of speech. We describe Stutter-TTS, an end-to-end neural text-to-speech model capable of synthesizing diverse types of stuttering utterances. We develop a simple, yet effective prosody-control strategy whereby additional tokens are introduced into source text during training to represent specific stuttering characteristics. By choosing the position of the stutter tokens, Stutter-TTS allows word-level control of where stuttering occurs in the synthesized utterance. We are able to synthesize stutter events with high accuracy (F1-scores between 0.63 and 0.84, depending on stutter type). By fine-tuning an ASR model on synthetic stuttered speech we are able to reduce word error by 5.7% relative on stuttered utterances, with only minor (<0.2% relative) degradation for fluent utterances.
Depth information is the foundation of perception, essential for autonomous driving, robotics, and other source-constrained applications. Promptly obtaining accurate and efficient depth information allows for a rapid response in dynamic environments. Sensor-based methods using LIDAR and RADAR obtain high precision at the cost of high power consumption, price, and volume. While due to advances in deep learning, vision-based approaches have recently received much attention and can overcome these drawbacks. In this work, we explore an extreme scenario in vision-based settings: estimate a depth map from one monocular image severely plagued by grid artifacts and blurry edges. To address this scenario, We first design a convolutional attention mechanism block (CAMB) which consists of channel attention and spatial attention sequentially and insert these CAMBs into skip connections. As a result, our novel approach can find the focus of current image with minimal overhead and avoid losses of depth features. Next, by combining the depth value, the gradients of X axis, Y axis and diagonal directions, and the structural similarity index measure (SSIM), we propose our novel loss function. Moreover, we utilize pixel blocks to accelerate the computation of the loss function. Finally, we show, through comprehensive experiments on two large-scale image datasets, i.e. KITTI and NYU-V2, that our method outperforms several representative baselines.
Compared with multi-class classification, multi-label classification that contains more than one class is more suitable in real life scenarios. Obtaining fully labeled high-quality datasets for multi-label classification problems, however, is extremely expensive, and sometimes even infeasible, with respect to annotation efforts, especially when the label spaces are too large. This motivates the research on partial-label classification, where only a limited number of labels are annotated and the others are missing. To address this problem, we first propose a pseudo-label based approach to reduce the cost of annotation without bringing additional complexity to the existing classification networks. Then we quantitatively study the impact of missing labels on the performance of classifier. Furthermore, by designing a novel loss function, we are able to relax the requirement that each instance must contain at least one positive label, which is commonly used in most existing approaches. Through comprehensive experiments on three large-scale multi-label image datasets, i.e. MS-COCO, NUS-WIDE, and Pascal VOC12, we show that our method can handle the imbalance between positive labels and negative labels, while still outperforming existing missing-label learning approaches in most cases, and in some cases even approaches with fully labeled datasets.
Conventional phrase grounding aims to localize noun phrases mentioned in a given caption to their corresponding image regions, which has achieved great success recently. Apparently, sole noun phrase grounding is not enough for cross-modal visual language understanding. Here we extend the task by considering pronouns as well. First, we construct a dataset of phrase grounding with both noun phrases and pronouns to image regions. Based on the dataset, we test the performance of phrase grounding by using a state-of-the-art literature model of this line. Then, we enhance the baseline grounding model with coreference information which should help our task potentially, modeling the coreference structures with graph convolutional networks. Experiments on our dataset, interestingly, show that pronouns are easier to ground than noun phrases, where the possible reason might be that these pronouns are much less ambiguous. Additionally, our final model with coreference information can significantly boost the grounding performance of both noun phrases and pronouns.
Multi-label image classification aims to predict all possible labels in an image. It is usually formulated as a partial-label learning problem, since it could be expensive in practice to annotate all the labels in every training image. Existing works on partial-label learning focus on the case where each training image is labeled with only a subset of its positive/negative labels. To effectively address partial-label classification, this paper proposes an end-to-end Generic Game-theoretic Network (G2NetPL) for partial-label learning, which can be applied to most partial-label settings, including a very challenging, but annotation-efficient case where only a subset of the training images are labeled, each with only one positive label, while the rest of the training images remain unlabeled. In G2NetPL, each unobserved label is associated with a soft pseudo label, which, together with the network, formulates a two-player non-zero-sum non-cooperative game. The objective of the network is to minimize the loss function with given pseudo labels, while the pseudo labels will seek convergence to 1 (positive) or 0 (negative) with a penalty of deviating from the predicted labels determined by the network. In addition, we introduce a confidence-aware scheduler into the loss of the network to adaptively perform easy-to-hard learning for different labels. Extensive experiments demonstrate that our proposed G2NetPL outperforms many state-of-the-art multi-label classification methods under various partial-label settings on three different datasets.
For medical cone-beam computed tomography (CBCT) imaging, the native receptor array of the flat-panel detector (FPD) is usually binned into a reduced matrix size. By doing so, the signal readout speed can be increased by over 4-9 times at the expense of sacrificing the spatial resolution by at least 50%-67%. Clearly, such tradition poses a main bottleneck in generating high spatial resolution and high temporal resolution CBCT images at the same time. In addition, the conventional FPD is also difficult in generating dual-energy CBCT images. In this paper, we propose an innovative super resolution dual-energy CBCT imaging method, named as suRi, based on dual-layer FPD (DL-FPD) to overcome these aforementioned difficulties at once. With suRi, specifically, an 1D or 2D sub-pixel (half pixel in this study) shifted binning is applied to replace the conventionally aligned binning to double the spatial sampling rate during the dual-energy data acquisition. As a result, the suRi approach provides a new strategy to enable high signal readout speed and high spatial resolution CBCT imaging with FPD. Moreover, a penalized likelihood material decomposition algorithm is developed to directly reconstruct the high resolution bases from the dual-energy CBCT projections containing spatial sub-pixel shifts. Experiments based on the single-layer FPD and DL-FPD are performed with physical phantoms and biological specimen to validate this newly developed suRi method. The synthesized monochromatic CT imaging results demonstrate that suRi can significantly improve the spatial image resolution by 46.15%. We believe the developed suRi method would be capable to greatly enhance the imaging performance of the DL-FPD based dual-energy CBCT systems in future.
Seismic full-waveform inversion (FWI) provides high resolution images of the subsurface by exploiting information in the recorded seismic waveforms. This is achieved by solving a highly nonnlinear and nonunique inverse problem. Bayesian inference is therefore used to quantify uncertainties in the solution. Variational inference is a method that provides probabilistic, Bayesian solutions efficiently using optimization. The method has been applied to 2D FWI problems to produce full Bayesian posterior distributions. However, due to higher dimensionality and more expensive computational cost, the performance of the method in 3D FWI problems remains unknown. We apply three variational inference methods to 3D FWI and analyse their performance. Specifically we apply automatic differential variational inference (ADVI), Stein variational gradient descent (SVGD) and stochastic SVGD (sSVGD), to a 3D FWI problem, and compare their results and computational cost. The results show that ADVI is the most computationally efficient method but systematically underestimates the uncertainty. The method can therefore be used to provide relatively rapid but approximate insights into the subsurface together with a lower bound estimate of the uncertainty. SVGD demands the highest computational cost, and still produces biased results. In contrast, by including a randomized term in the SVGD dynamics, sSVGD becomes a Markov chain Monte Carlo method and provides the most accurate results at intermediate computational cost. We thus conclude that 3D variational full-waveform inversion is practically applicable, at least in small problems, and can be used to image the Earth's interior and to provide reasonable uncertainty estimates on those images.
An incidence of a graph $G$ is a vertex-edge pair $(v,e)$ such that $v$ is incidence with $e$. A conflict-free incidence coloring of a graph is a coloring of the incidences in such a way that two incidences $(u,e)$ and $(v,f)$ get distinct colors if and only if they conflict each other, i.e.,(i) $u=v$, (ii) $uv$ is $e$ or $f$, or (iii) there is a vertex $w$ such that $uw=e$ and $vw=f$. The minimum number of colors used among all conflict-free incidence colorings of a graph is the conflict-free incidence chromatic number. A graph is outer-1-planar if it can be drawn in the plane so that vertices are on the outer-boundary and each edge is crossed at most once. In this paper, we show that the conflict-free incidence chromatic number of an outer-1-planar graph with maximum degree $Δ$ is either $2Δ$ or $2Δ+1$ unless the graph is a cycle on three vertices, and moreover, all outer-1-planar graphs with conflict-free incidence chromatic number $2Δ$ or $2Δ+1$ are completely characterized. An efficient algorithm for constructing an optimal conflict-free incidence coloring of a connected outer-1-planar graph is given.
For weakly supervised anomaly detection, most existing work is limited to the problem of inadequate video representation due to the inability of modeling long-term contextual information. To solve this, we propose a novel weakly supervised adaptive graph convolutional network (WAGCN) to model the complex contextual relationship among video segments. By which, we fully consider the influence of other video segments on the current one when generating the anomaly probability score for each segment. Firstly, we combine the temporal consistency as well as feature similarity of video segments to construct a global graph, which makes full use of the association information among spatial-temporal features of anomalous events in videos. Secondly, we propose a graph learning layer in order to break the limitation of setting topology manually, which can extract graph adjacency matrix based on data adaptively and effectively. Extensive experiments on two public datasets (i.e., UCF-Crime dataset and ShanghaiTech dataset) demonstrate the effectiveness of our approach which achieves state-of-the-art performance.
In this paper, we show existence and uniqueness of solutions of the infinite horizon McKean-Vlasov FBSDEs using two different methods, which lead to two different sets of assumptions. We use these results to solve the infinite horizon mean field type control problems and mean field games.
In recent years, decentralized bilevel optimization problems have received increasing attention in the networking and machine learning communities thanks to their versatility in modeling decentralized learning problems over peer-to-peer networks (e.g., multi-agent meta-learning, multi-agent reinforcement learning, personalized training, and Byzantine-resilient learning). However, for decentralized bilevel optimization over peer-to-peer networks with limited computation and communication capabilities, how to achieve low sample and communication complexities are two fundamental challenges that remain under-explored so far. In this paper, we make the first attempt to investigate the class of decentralized bilevel optimization problems with nonconvex and strongly-convex structure corresponding to the outer and inner subproblems, respectively. Our main contributions in this paper are two-fold: i) We first propose a deterministic algorithm called INTERACT (inner-gradient-descent-outer-tracked-gradient) that requires the sample complexity of $\mathcal{O}(n ε^{-1})$ and communication complexity of $\mathcal{O}(ε^{-1})$ to solve the bilevel optimization problem, where $n$ and $ε> 0$ are the number of samples at each agent and the desired stationarity gap, respectively. ii) To relax the need for full gradient evaluations in each iteration, we propose a stochastic variance-reduced version of INTERACT (SVR-INTERACT), which improves the sample complexity to $\mathcal{O}(\sqrt{n} ε^{-1})$ while achieving the same communication complexity as the deterministic algorithm. To our knowledge, this work is the first that achieves both low sample and communication complexities for solving decentralized bilevel optimization problems over networks. Our numerical experiments also corroborate our theoretical findings.
High-resolution dynamic X-ray detector is crucial for time-resolved digital radiography (DR) imaging and fast 3D medical computed tomography (CT) imaging. Recently, perovskites have become promising alternatives to conventional semi-conductor materials, e.g., Si, a-Se and CdTe, for direct X-ray detection. However, the feasibility of their combination with high-speed pixelated complementary metal-oxide-semiconductor (CMOS) arrays remains unknown. This work originally reports an innovative direct-conversion X-ray detector fabricated with 300 micrometer thick inorganic perovskite film printed on a tailored CMOS array. In-house measurements demonstrate that the CsPbBr3 film has excellent optoelectric properties of an electron mobility-lifetime product of 3.40x10$^{-5}$ cm$^2$ V$^{-1}$, and the X-ray detector exhibits high sensitivity of 9341uC Gy$_{\rm air}^{-1}$ cm$^{-2}$, and low detection limit of 588 nGy$_{\rm air}^{-1}$. This CMOS X-ray imaging detector achieves a high spatial resolution up to 5.5 lp/mm (close to the resolution limit of 6.0 lp/mm), and >300 frame per second (fps) readout speed. DR image of a resolution pattern phantom and a anesthesia mice, CT images of a biological specimen are acquired for the first time.
Successful Machine Learning based Named Entity Recognition models could fail on texts from some special domains, for instance, Chinese addresses and e-commerce titles, where requires adequate background knowledge. Such texts are also difficult for human annotators. In fact, we can obtain some potentially helpful information from correlated texts, which have some common entities, to help the text understanding. Then, one can easily reason out the correct answer by referencing correlated samples. In this paper, we suggest enhancing NER models with correlated samples. We draw correlated samples by the sparse BM25 retriever from large-scale in-domain unlabeled data. To explicitly simulate the human reasoning process, we perform a training-free entity type calibrating by majority voting. To capture correlation features in the training stage, we suggest to model correlated samples by the transformer-based multi-instance cross-encoder. Empirical results on datasets of the above two domains show the efficacy of our methods.
Although the cosmic curvature has been tightly constrained in the standard cosmological model using observations of cosmic microwave background anisotropies, it is still of great importance to independently measure this key parameter using only late-universe observations in a cosmological model-independent way. The distance sum rule in strong gravitational lensing (SGL) provides such a way, provided that the three distances in the sum rule can be calibrated by other observations. In this paper, we propose that gravitational waves (GWs) can be used to provide the distance calibration in the SGL method, which can avoid the dependence on distance ladder and cover a wider redshift range. Using the simulated GW standard siren observation by the Einstein Telescope as an example, we show that this scheme is feasible and advantageous. We find that $ΔΩ_k\simeq 0.17$ with the current SGL data, which is slightly more precise than the case of using SN to calibrate. Furthermore, we consider the forthcoming LSST survey that is expected to observe many SGL systems, and we find that about $10^4$ SGL data could provide the precise measurement of $ΔΩ_k\simeq 10^{-2}$ with the help of GWs. In addition, our results confirm that this method of constraining $Ω_k$ is strongly dependent on lens models. However, obtaining a more accurate phenomenological model for lens galaxies is highly predictable as future massive surveys observe more and more SGL samples, which will significantly improve the constraint of cosmic curvature.
Non-line-of-sight (NLOS) imaging is an emerging technique for detecting objects behind obstacles or around corners. Recent studies on passive NLOS mainly focus on steady-state measurement and reconstruction methods, which show limitations in recognition of moving targets. To the best of our knowledge, we propose a novel event-based passive NLOS imaging method. We acquire asynchronous event-based data which contains detailed dynamic information of the NLOS target, and efficiently ease the degradation of speckle caused by movement. Besides, we create the first event-based NLOS imaging dataset, NLOS-ES, and the event-based feature is extracted by time-surface representation. We compare the reconstructions through event-based data with frame-based data. The event-based method performs well on PSNR and LPIPS, which is 20% and 10% better than frame-based method, while the data volume takes only 2% of traditional method.
Histopathology whole slide images (WSIs) play a very important role in clinical studies and serve as the gold standard for many cancer diagnoses. However, generating automatic tools for processing WSIs is challenging due to their enormous sizes. Currently, to deal with this issue, conventional methods rely on a multiple instance learning (MIL) strategy to process a WSI at patch level. Although effective, such methods are computationally expensive, because tiling a WSI into patches takes time and does not explore the spatial relations between these tiles. To tackle these limitations, we propose a locally supervised learning framework which processes the entire slide by exploring the entire local and global information that it contains. This framework divides a pre-trained network into several modules and optimizes each module locally using an auxiliary model. We also introduce a random feature reconstruction unit (RFR) to preserve distinguishing features during training and improve the performance of our method by 1% to 3%. Extensive experiments on three publicly available WSI datasets: TCGA-NSCLC, TCGA-RCC and LKS, highlight the superiority of our method on different classification tasks. Our method outperforms the state-of-the-art MIL methods by 2% to 5% in accuracy, while being 7 to 10 times faster. Additionally, when dividing it into eight modules, our method requires as little as 20% of the total gpu memory required by end-to-end training. Our code is available at https://github.com/cvlab-stonybrook/local_learning_wsi.