arXiv subjects

Yu Zhang

Publications and source records attributed to Yu Zhang.

At least 1,225 records · Page 68Linked to original sources

arXiv2021

Dynamic Feature Alignment for Semi-supervised Domain Adaptation

Most research on domain adaptation has focused on the purely unsupervised setting, where no labeled examples in the target domain are available. However, in many real-world scenarios, a small amount of labeled target data is available and can be used to improve adaptation. We address this semi-supervised setting and propose to use dynamic feature alignment to address both inter- and intra-domain discrepancy. Unlike previous approaches, which attempt to align source and target features within a mini-batch, we propose to align the target features to a set of dynamically updated class prototypes, which we use both for minimizing divergence and pseudo-labeling. By updating based on class prototypes, we avoid problems that arise in previous approaches due to class imbalances. Our approach, which doesn't require extensive tuning or adversarial training, significantly improves the state of the art for semi-supervised domain adaptation. We provide a quantitative evaluation on two standard datasets, DomainNet and Office-Home, and performance analysis.

cs.CV↗

arXiv2021

Generation of Intense Phase-Stable Femtosecond Hard X-ray Pulse Pairs

Coherent nonlinear spectroscopies and imaging in the X-ray domain provide direct insight into the coupled motions of electrons and nuclei with resolution on the electronic length and time scale. The experimental realization of such techniques will strongly benefit from access to intense, coherent pairs of femtosecond X-ray pulses. We have observed phase-stable X-ray pulse pairs containing more thank 3 x 10e7 photons at 5.9 keV (2.1 Angstrom) with about 1 fs duration and 2-5 fs separation. The highly directional pulse pairs are manifested by interference fringes in the superfluorescent and seeded stimulated manganese K-alpha emission induced by an X-ray free-electron laser. The fringes constitute the time-frequency X-ray analogue of the Young double-slit interference allowing for frequency-domain X-ray measurements with attosecond time resolution.

physics.atom-ph↗

arXiv2021

Region Semantically Aligned Network for Zero-Shot Learning

Zero-shot learning (ZSL) aims to recognize unseen classes based on the knowledge of seen classes. Previous methods focused on learning direct embeddings from global features to the semantic space in hope of knowledge transfer from seen classes to unseen classes. However, an unseen class shares local visual features with a set of seen classes and leveraging global visual features makes the knowledge transfer ineffective. To tackle this problem, we propose a Region Semantically Aligned Network (RSAN), which maps local features of unseen classes to their semantic attributes. Instead of using global features which are obtained by an average pooling layer after an image encoder, we directly utilize the output of the image encoder which maintains local information of the image. Concretely, we obtain each attribute from a specific region of the output and exploit these attributes for recognition. As a result, the knowledge of seen classes can be successfully transferred to unseen classes in a region-bases manner. In addition, we regularize the image encoder through attribute regression with a semantic knowledge to extract robust and attribute-related visual features. Experiments on several standard ZSL datasets reveal the benefit of the proposed RSAN method, outperforming state-of-the-art methods.

cs.CV↗

arXiv2021

Crossover behavior in the magnetoresistance of thin flakes of the topological material ZrTe5

ZrTe5 is a layered material that exhibits intricate topological effects. Intensive theoretically and experimental efforts have been devoted to try to understand the physics in this materials. In this paper the temperature dependent magneto-transport properties of ZrTe5 thin flakes are investigated. A characteristic temperature T* is observed in the temperature dependence of three different types of magnetoresistance simultaneously, which are the saturated Hall anomaly, the chiral anomaly and the longitudinal magnetoresistance. Furthermore, the value of T* decreases monotonically from 200K to 160K with increasing thickness of the ZrTe5 thin flakes from 42nm to 89nm. Temperature induced topological phase transitions are attributed to the cause of such anomaly in the three types of magnetoresistance at T*. Our findings provide a multi-parameter indicator for the emergence of topological phase transition in ZrTe5 and could be extended to the study of other topological materials. The temperature dependence of the three types of magnetoresistance also shed light on the role of anomalous Hall Effect in the transport properties of ZrTe5.

cond-mat.mtrl-sci↗

arXiv2021

Learning Reflection Beamforming Codebooks for Arbitrary RIS and Non-Stationary Channels

Reconfigurable intelligent surfaces (RIS) are expected to play an important role in future wireless communication systems. These surfaces typically rely on their reflection beamforming codebooks to reflect and focus the signal on the target receivers. Prior work has mainly considered pre-defined RIS beamsteering codebooks that do not adapt to the environment and hardware and lead to large beam training overhead. In this work, a novel deep reinforcement learning based framework is developed to efficiently construct the RIS reflection beam codebook. This framework adopts a multi-level design approach that transfers the learning between the multiple RIS subarrays, which speeds up the learning convergence and highly reduces the computational complexity for large RIS surfaces. The proposed approach is generic for co-located/distributed RIS surfaces with arbitrary array geometries and with stationary/non-stationary channels. Further, the developed solution does not require explicitly channel knowledge and adapts the codebook beams to the surrounding environment, user distribution, and hardware characteristics. Simulation results show that the proposed learning framework can learn optimized interaction codebooks within reasonable iterations. Besides, with only 6 beams, the learned codebook outperforms a 256-beam DFT codebook, which significantly reduces the beam training overhead.

cs.IT↗

arXiv2021

Electrically Switchable van der Waals Magnon Valves

Van der Waals magnets have emerged as a fertile ground for the exploration of highly tunable spin physics and spin-related technology. Two-dimensional (2D) magnons in van der Waals magnets are collective excitation of spins under strong confinement. Although considerable progress has been made in understanding 2D magnons, a crucial magnon device called the van der Waals magnon valve, in which the magnon signal can be completely and repeatedly turned on and off electrically, has yet to be realized. Here we demonstrate such magnon valves based on van der Waals antiferromagnetic insulator MnPS3. By applying DC electric current through the gate electrode, we show that the second harmonic thermal magnon (SHM) signal can be tuned from positive to negative. The guaranteed zero crossing during this tuning demonstrates a complete blocking of SHM transmission, arising from the nonlinear gate dependence of the non-equilibrium magnon density in the 2D spin channel. Using the switchable magnon valves we demonstrate a magnon-based inverter. These results illustrate the potential of van der Waals anti-ferromagnets for studying highly tunable spin-wave physics and for application in magnon-base circuitry in future information technology.

cond-mat.mtrl-sci↗

arXiv2021

Comparison of Object Detection Algorithms Using Video and Thermal Images Collected from a UAS Platform: An Application of Drones in Traffic Management

There is a rapid growth of applications of Unmanned Aerial Vehicles (UAVs) in traffic management, such as traffic surveillance, monitoring, and incident detection. However, the existing literature lacks solutions to real-time incident detection while addressing privacy issues in practice. This study explored real-time vehicle detection algorithms on both visual and infrared cameras and conducted experiments comparing their performance. Red Green Blue (RGB) videos and thermal images were collected from a UAS platform along highways in the Tampa, Florida, area. Experiments were designed to quantify the performance of a real-time background subtraction-based method in vehicle detection from a stationary camera on hovering UAVs under free-flow conditions. Several parameters were set in the experiments based on the geometry of the drone and sensor relative to the roadway. The results show that a background subtraction-based method can achieve good detection performance on RGB images (F1 scores around 0.9 for most cases), and a more varied performance is seen on thermal images with different azimuth angles. The results of these experiments will help inform the development of protocols, standards, and guidance for the use of drones to detect highway congestion and provide input for the development of incident detection algorithms.

cs.CV↗

arXiv2021

A Simple Self-calibration Method for The Internal Time Synchronization of MEMS LiDAR

This paper proposes a simple self-calibration method for the internal time synchronization of MEMS(Micro-electromechanical systems) LiDAR during research and development. Firstly, we introduced the problem of internal time misalignment in MEMS lidar. Then, a robust Minimum Vertical Gradient(MVG) prior is proposed to calibrate the time difference between the laser and MEMS mirror, which can be calculated automatically without any artificial participation or specially designed cooperation target. Finally, actual experiments on MEMS LiDARs are implemented to demonstrate the effectiveness of the proposed method. It should be noted that the calibration can be implemented in a simple laboratory environment without any ranging equipment and artificial participation, which greatly accelerate the progress of research and development in practical applications.

cs.CV↗

arXiv2021

Generating Active Explicable Plans in Human-Robot Teaming

Intelligent robots are redefining a multitude of critical domains but are still far from being fully capable of assisting human peers in day-to-day tasks. An important requirement of collaboration is for each teammate to maintain and respect an understanding of the others' expectations of itself. Lack of which may lead to serious issues such as loose coordination between teammates, reduced situation awareness, and ultimately teaming failures. Hence, it is important for robots to behave explicably by meeting the human's expectations. One of the challenges here is that the expectations of the human are often hidden and can change dynamically as the human interacts with the robot. However, existing approaches to generating explicable plans often assume that the human's expectations are known and static. In this paper, we propose the idea of active explicable planning to relax this assumption. We apply a Bayesian approach to model and predict dynamic human belief and expectations to make explicable planning more anticipatory. We hypothesize that active explicable plans can be more efficient and explicable at the same time, when compared to explicable plans generated by the existing methods. In our experimental evaluation, we verify that our approach generates more efficient explicable plans while successfully capturing the dynamic belief change of the human teammate.

cs.RO↗

arXiv2021

Logic-level Evidence Retrieval and Graph-based Verification Network for Table-based Fact Verification

Table-based fact verification task aims to verify whether the given statement is supported by the given semi-structured table. Symbolic reasoning with logical operations plays a crucial role in this task. Existing methods leverage programs that contain rich logical information to enhance the verification process. However, due to the lack of fully supervised signals in the program generation process, spurious programs can be derived and employed, which leads to the inability of the model to catch helpful logical operations. To address the aforementioned problems, in this work, we formulate the table-based fact verification task as an evidence retrieval and reasoning framework, proposing the Logic-level Evidence Retrieval and Graph-based Verification network (LERGV). Specifically, we first retrieve logic-level program-like evidence from the given table and statement as supplementary evidence for the table. After that, we construct a logic-level graph to capture the logical relations between entities and functions in the retrieved evidence, and design a graph-based verification network to perform logic-level graph-based reasoning based on the constructed graph to classify the final entailment relation. Experimental results on the large-scale benchmark TABFACT show the effectiveness of the proposed approach.

cs.AI↗

arXiv2021

W2v-BERT: Combining Contrastive Learning and Masked Language Modeling for Self-Supervised Speech Pre-Training

Motivated by the success of masked language modeling~(MLM) in pre-training natural language processing models, we propose w2v-BERT that explores MLM for self-supervised speech representation learning. w2v-BERT is a framework that combines contrastive learning and MLM, where the former trains the model to discretize input continuous speech signals into a finite set of discriminative speech tokens, and the latter trains the model to learn contextualized speech representations via solving a masked prediction task consuming the discretized tokens. In contrast to existing MLM-based speech pre-training frameworks such as HuBERT, which relies on an iterative re-clustering and re-training process, or vq-wav2vec, which concatenates two separately trained modules, w2v-BERT can be optimized in an end-to-end fashion by solving the two self-supervised tasks~(the contrastive task and MLM) simultaneously. Our experiments show that w2v-BERT achieves competitive results compared to current state-of-the-art pre-trained models on the LibriSpeech benchmarks when using the Libri-Light~60k corpus as the unsupervised data. In particular, when compared to published models such as conformer-based wav2vec~2.0 and HuBERT, our model shows~5\% to~10\% relative WER reduction on the test-clean and test-other subsets. When applied to the Google's Voice Search traffic dataset, w2v-BERT outperforms our internal conformer-based wav2vec~2.0 by more than~30\% relatively.

cs.LG↗

arXiv2021

Multi-source Heterogeneous Domain Adaptation with Conditional Weighting Adversarial Network

Heterogeneous domain adaptation (HDA) tackles the learning of cross-domain samples with both different probability distributions and feature representations. Most of the existing HDA studies focus on the single-source scenario. In reality, however, it is not uncommon to obtain samples from multiple heterogeneous domains. In this article, we study the multisource HDA problem and propose a conditional weighting adversarial network (CWAN) to address it. The proposed CWAN adversarially learns a feature transformer, a label classifier, and a domain discriminator. To quantify the importance of different source domains, CWAN introduces a sophisticated conditional weighting scheme to calculate the weights of the source domains according to the conditional distribution divergence between the source and target domains. Different from existing weighting schemes, the proposed conditional weighting scheme not only weights the source domains but also implicitly aligns the conditional distributions during the optimization process. Experimental results clearly demonstrate that the proposed CWAN performs much better than several state-of-the-art methods on four real-world datasets.

cs.LG↗

arXiv2021

Domain Adaptation by Maximizing Population Correlation with Neural Architecture Search

In Domain Adaptation (DA), where the feature distributions of the source and target domains are different, various distance-based methods have been proposed to minimize the discrepancy between the source and target domains to handle the domain shift. In this paper, we propose a new similarity function, which is called Population Correlation (PC), to measure the domain discrepancy for DA. Base on the PC function, we propose a new method called Domain Adaptation by Maximizing Population Correlation (DAMPC) to learn a domain-invariant feature representation for DA. Moreover, most existing DA methods use hand-crafted bottleneck networks, which may limit the capacity and flexibility of the corresponding model. Therefore, we further propose a method called DAMPC with Neural Architecture Search (DAMPC-NAS) to search the optimal network architecture for DAMPC. Experiments on several benchmark datasets, including Office-31, Office-Home, and VisDA-2017, show that the proposed DAMPC-NAS method achieves better results than state-of-the-art DA methods.

cs.CV↗

arXiv2021

Scaling End-to-End Models for Large-Scale Multilingual ASR

Building ASR models across many languages is a challenging multi-task learning problem due to large variations and heavily unbalanced data. Existing work has shown positive transfer from high resource to low resource languages. However, degradations on high resource languages are commonly observed due to interference from the heterogeneous multilingual data and reduction in per-language capacity. We conduct a capacity study on a 15-language task, with the amount of data per language varying from 7.6K to 53.5K hours. We adopt GShard [1] to efficiently scale up to 10B parameters. Empirically, we find that (1) scaling the number of model parameters is an effective way to solve the capacity bottleneck - our 500M-param model already outperforms monolingual baselines and scaling it to 1B and 10B brought further quality gains; (2) larger models are not only more data efficient, but also more efficient in terms of training cost as measured in TPU days - the 1B-param model reaches the same accuracy at 34% of training time as the 500M-param model; (3) given a fixed capacity budget, adding depth works better than width and large encoders do better than large decoders; (4) with continuous training, they can be adapted to new languages and domains.

cs.CL↗

arXiv2021

Pushing the Limits of Non-Autoregressive Speech Recognition

We combine recent advancements in end-to-end speech recognition to non-autoregressive automatic speech recognition. We push the limits of non-autoregressive state-of-the-art results for multiple datasets: LibriSpeech, Fisher+Switchboard and Wall Street Journal. Key to our recipe, we leverage CTC on giant Conformer neural network architectures with SpecAugment and wav2vec2 pre-training. We achieve 1.8%/3.6% WER on LibriSpeech test/test-other sets, 5.1%/9.8% WER on Switchboard, and 3.4% on the Wall Street Journal, all without a language model.

eess.AS↗

arXiv2021

Distantly-Supervised Named Entity Recognition with Noise-Robust Learning and Language Model Augmented Self-Training

We study the problem of training named entity recognition (NER) models using only distantly-labeled data, which can be automatically obtained by matching entity mentions in the raw text with entity types in a knowledge base. The biggest challenge of distantly-supervised NER is that the distant supervision may induce incomplete and noisy labels, rendering the straightforward application of supervised learning ineffective. In this paper, we propose (1) a noise-robust learning scheme comprised of a new loss function and a noisy label removal step, for training NER models on distantly-labeled data, and (2) a self-training method that uses contextualized augmentations created by pre-trained language models to improve the generalization ability of the NER model. On three benchmark datasets, our method achieves superior performance, outperforming existing distantly-supervised NER models by significant margins.

cs.CL↗

arXiv2021

Contrastive Cross-Modal Pre-Training: A General Strategy for Small Sample Medical Imaging

A key challenge in training neural networks for a given medical imaging task is often the difficulty of obtaining a sufficient number of manually labeled examples. In contrast, textual imaging reports, which are often readily available in medical records, contain rich but unstructured interpretations written by experts as part of standard clinical practice. We propose using these textual reports as a form of weak supervision to improve the image interpretation performance of a neural network without requiring additional manually labeled examples. We use an image-text matching task to train a feature extractor and then fine-tune it in a transfer learning setting for a supervised task using a small labeled dataset. The end result is a neural network that automatically interprets imagery without requiring textual reports during inference. This approach can be applied to any task for which text-image pairs are readily available. We evaluate our method on three classification tasks and find consistent performance improvements, reducing the need for labeled data by 67%-98%.

cs.LG↗

arXiv2021

Spin-dependent metalens with intensity-adjustable dual-focused vortex beams

Vortex beams with orbital angular momentum has been attracting tremendous attention due to their considerable applications ranging from optical tweezers to quantum information processing. Metalens, an ultra-compact and multifunctional device, provide a desired platform for designing vortex beams. A spin-dependent metalens can boost the freedom to further satisfy practical applications. By combining geometric phase and propagation phase, we propose and demonstrate an approach to design a spin-dependent metalens generating dual-focused vortex beams along longitudinal or transverse direction, i.e., metalenses with predesigned spin-dependent phase profiles. Under the illumination of an elliptical polarization incident beam, two spin-dependent focused vortex beams can be observed, and the relative focal intensity of them can be easily adjusted by modulating the ellipticity of the incident beam. Moreover, we also demonstrated that the separate distance between these dual-focused beams and their topological charges could be simultaneously tailored at will, which may have a profound impact on optical trapping and manipulation in photonics.

physics.optics↗