arXiv subjects

Zhiqiang Shen

Publications and source records attributed to Zhiqiang Shen.

At least 163 records · Page 9Linked to original sources

arXiv2022

Multi-frequency VLBI Observations of the M 84 Inner Jet/Counterjet

Observational studies of inner-most regions of the edge-on jets in nearby active galactic nuclei (AGN) are crucial to understand their kinematics and morphology. For the inner jet of the nearby low luminosity AGN in M 84, we present new high-sensitivity observations with very long baseline interferometry since 2019, as well as archival Very Long Baseline Array observations in 2014. We find that the compact core in M 84 has an inverted-to-flat spectrum from 1.5 to 88 GHz. Based on the turnover frequency of $4.2\pm 0.2$ GHz in the spectrum, we estimated a magnetic field strength of 1-10mG and an electron number density of $\sim 10^5 cm^{-3}$ in the core region. Three inner jet components within $\sim 3$ mas from the core are identified and traced in the images at 22 GHz, whose apparent speeds are 0.11 c, 0.27 c, and 0.32 c, respectively. We calculate the viewing angle of $\sim58$ degree for the inner jet based on the proper motion and the flux ratio of jet-to-counterjet. A propagating sinusoidal model with a wavelength of $\sim 3.4$ mas is used to fit the helical morphology of the jet extended to 20 mas ($\sim 2.2\times 10^4$ Schwarzschild Radii).

astro-ph.HE↗

arXiv2022

Exploring Simple and Transferable Recognition-Aware Image Processing

Recent progress in image recognition has stimulated the deployment of vision systems at an unprecedented scale. As a result, visual data are now often consumed not only by humans but also by machines. Existing image processing methods only optimize for better human perception, yet the resulting images may not be accurately recognized by machines. This can be undesirable, e.g., the images can be improperly handled by search engines or recommendation systems. In this work, we examine simple approaches to improve machine recognition of processed images: optimizing the recognition loss directly on the image processing network or through an intermediate input transformation model. Interestingly, the processing model's ability to enhance recognition quality can transfer when evaluated on models of different architectures, recognized categories, tasks and training datasets. This makes the methods applicable even when we do not have the knowledge of future recognition models, e.g., when uploading processed images to the Internet. We conduct experiments on multiple image processing tasks paired with ImageNet classification and PASCAL VOC detection as recognition tasks. With these simple yet effective methods, substantial accuracy gain can be achieved with strong transferability and minimal image quality loss. Through a user study we further show that the accuracy gain can transfer to a black-box cloud model. Finally, we try to explain this transferability phenomenon by demonstrating the similarities of different models' decision boundaries. Code is available at https://github.com/liuzhuang13/Transferable_RA .

cs.CV↗

arXiv2022

A Q-band line survey towards Orion KL using the Tianma radio telescope

We have conducted a line survey towards Orion KL using the Q-band receiver of Tianma 65 m radio telescope (TMRT), covering 34.8--50 GHz with a velocity resolution between 0.79 km s$^{-1}$ and 0.55 km s$^{-1}$ respectively. The observations reach a sensitivity on the level of 1-8 mK, proving that the TMRT is sensitive for conducting deep line surveys. In total, 597 Gaussian features are extracted. Among them, 177 radio recombination lines (RRLs) are identified, including 126, 40 and 11 RRLs of hydrogen, helium and carbon, with a maximum $Δn$ of 16, 7, and 3, respectively. The carbon RRLs are confirmed to originate from photodissociation regions with a $V_{\rm LSR}\sim$9 km s$^{-1}$. In addition, 371 molecular transitions of 53 molecular species are identified. Twenty-one molecular species of this survey were not firmly detected in the Q band by Rizzo et al. (2017), including species such as H$_2$CS, HCOOH, C$_2$H$_5$OH, H$_2^{13}$CO, H$_2$CCO, CH$_3$CHO, CH$_2$OCH$_2$, HCN $v_2=1$, and CH$_3$OCHO $v_t=1$. In particular, the vibrationally excited states of ethyl cyanide (C$_2$H$_5$CN $v$13/$v$21) are for the first time firmly detected in the Q band. NH$_3$ (15,15) and (16,16) are identified, and they are so far the highest transitions of the NH$_3$ inversion lines detected towards Orion KL. All the identified lines can be reproduced by a radiative transfer model.

astro-ph.GA↗

arXiv2022

Sliced Recursive Transformer

We present a neat yet effective recursive operation on vision transformers that can improve parameter utilization without involving additional parameters. This is achieved by sharing weights across the depth of transformer networks. The proposed method can obtain a substantial gain (~2%) simply using naive recursive operation, requires no special or sophisticated knowledge for designing principles of networks, and introduces minimal computational overhead to the training procedure. To reduce the additional computation caused by recursive operation while maintaining the superior accuracy, we propose an approximating method through multiple sliced group self-attentions across recursive layers which can reduce the cost consumption by 10~30% with minimal performance loss. We call our model Sliced Recursive Transformer (SReT), a novel and parameter-efficient vision transformer design that is compatible with a broad range of other designs for efficient ViT architectures. Our best model establishes significant improvement on ImageNet-1K over state-of-the-art methods while containing fewer parameters. The proposed weight sharing mechanism by sliced recursion structure allows us to build a transformer with more than 100 or even 1000 shared layers with ease while keeping a compact size (13~15M), to avoid optimization difficulties when the model is too large. The flexible scalability has shown great potential for scaling up models and constructing extremely deep vision transformers. Code is available at https://github.com/szq0214/SReT.

cs.CV↗

arXiv2022

Data-Free Neural Architecture Search via Recursive Label Calibration

This paper aims to explore the feasibility of neural architecture search (NAS) given only a pre-trained model without using any original training data. This is an important circumstance for privacy protection, bias avoidance, etc., in real-world scenarios. To achieve this, we start by synthesizing usable data through recovering the knowledge from a pre-trained deep neural network. Then we use the synthesized data and their predicted soft-labels to guide neural architecture search. We identify that the NAS task requires the synthesized data (we target at image domain here) with enough semantics, diversity, and a minimal domain gap from the natural images. For semantics, we propose recursive label calibration to produce more informative outputs. For diversity, we propose a regional update strategy to generate more diverse and semantically-enriched synthetic data. For minimal domain gap, we use input and feature-level regularization to mimic the original data distribution in latent space. We instantiate our proposed framework with three popular NAS algorithms: DARTS, ProxylessNAS and SPOS. Surprisingly, our results demonstrate that the architectures discovered by searching with our synthetic data achieve accuracy that is comparable to, or even higher than, architectures discovered by searching from the original ones, for the first time, deriving the conclusion that NAS can be done effectively with no need of access to the original or called natural data if the synthesis method is well designed.

cs.LG↗

arXiv2022

SDQ: Stochastic Differentiable Quantization with Mixed Precision

In order to deploy deep models in a computationally efficient manner, model quantization approaches have been frequently used. In addition, as new hardware that supports mixed bitwidth arithmetic operations, recent research on mixed precision quantization (MPQ) begins to fully leverage the capacity of representation by searching optimized bitwidths for different layers and modules in a network. However, previous studies mainly search the MPQ strategy in a costly scheme using reinforcement learning, neural architecture search, etc., or simply utilize partial prior knowledge for bitwidth assignment, which might be biased and sub-optimal. In this work, we present a novel Stochastic Differentiable Quantization (SDQ) method that can automatically learn the MPQ strategy in a more flexible and globally-optimized space with smoother gradient approximation. Particularly, Differentiable Bitwidth Parameters (DBPs) are employed as the probability factors in stochastic quantization between adjacent bitwidth choices. After the optimal MPQ strategy is acquired, we further train our network with entropy-aware bin regularization and knowledge distillation. We extensively evaluate our method for several networks on different hardware (GPUs and FPGA) and datasets. SDQ outperforms all state-of-the-art mixed or single precision quantization with a lower bitwidth and is even better than the full-precision counterparts across various ResNet and MobileNet families, demonstrating the effectiveness and superiority of our method.

cs.LG↗

arXiv2022

Vision Transformer Slimming: Multi-Dimension Searching in Continuous Optimization Space

This paper explores the feasibility of finding an optimal sub-model from a vision transformer and introduces a pure vision transformer slimming (ViT-Slim) framework. It can search a sub-structure from the original model end-to-end across multiple dimensions, including the input tokens, MHSA and MLP modules with state-of-the-art performance. Our method is based on a learnable and unified $\ell_1$ sparsity constraint with pre-defined factors to reflect the global importance in the continuous searching space of different dimensions. The searching process is highly efficient through a single-shot training scheme. For instance, on DeiT-S, ViT-Slim only takes ~43 GPU hours for the searching process, and the searched structure is flexible with diverse dimensionalities in different modules. Then, a budget threshold is employed according to the requirements of accuracy-FLOPs trade-off on running devices, and a re-training process is performed to obtain the final model. The extensive experiments show that our ViT-Slim can compress up to 40% of parameters and 40% FLOPs on various vision transformers while increasing the accuracy by ~0.6% on ImageNet. We also demonstrate the advantage of our searched models on several downstream datasets. Our code is available at https://github.com/Arnav0400/ViT-Slim.

cs.CV↗

arXiv2022

Nonuniform-to-Uniform Quantization: Towards Accurate Quantization via Generalized Straight-Through Estimation

The nonuniform quantization strategy for compressing neural networks usually achieves better performance than its counterpart, i.e., uniform strategy, due to its superior representational capacity. However, many nonuniform quantization methods overlook the complicated projection process in implementing the nonuniformly quantized weights/activations, which incurs non-negligible time and space overhead in hardware deployment. In this study, we propose Nonuniform-to-Uniform Quantization (N2UQ), a method that can maintain the strong representation ability of nonuniform methods while being hardware-friendly and efficient as the uniform quantization for model inference. We achieve this through learning the flexible in-equidistant input thresholds to better fit the underlying distribution while quantizing these real-valued inputs into equidistant output levels. To train the quantized network with learnable input thresholds, we introduce a generalized straight-through estimator (G-STE) for intractable backward derivative calculation w.r.t. threshold parameters. Additionally, we consider entropy preserving regularization to further reduce information loss in weight quantization. Even under this adverse constraint of imposing uniformly quantized weights and activations, our N2UQ outperforms state-of-the-art nonuniform quantization methods by 0.5~1.7 on ImageNet, demonstrating the contribution of N2UQ design. Code and models are available at: https://github.com/liuzechun/Nonuniform-to-Uniform-Quantization.

cs.CV↗

arXiv2022

Stereo Neural Vernier Caliper

We propose a new object-centric framework for learning-based stereo 3D object detection. Previous studies build scene-centric representations that do not consider the significant variation among outdoor instances and thus lack the flexibility and functionalities that an instance-level model can offer. We build such an instance-level model by formulating and tackling a local update problem, i.e., how to predict a refined update given an initial 3D cuboid guess. We demonstrate how solving this problem can complement scene-centric approaches in (i) building a coarse-to-fine multi-resolution system, (ii) performing model-agnostic object location refinement, and (iii) conducting stereo 3D tracking-by-detection. Extensive experiments demonstrate the effectiveness of our approach, which achieves state-of-the-art performance on the KITTI benchmark. Code and pre-trained models are available at https://github.com/Nicholasli1995/SNVC.

cs.CV↗

arXiv2022

Radio properties of the OH megamaser galaxy IIZw 096

Based on the two epochs EVN archive data from OH line observations of IIZw 096, we confirm that the high-resolution OH emission in this source mainly comes from two spots (OH1 and OH2) of comp D1 of this merging system. We found no significant variations in the OH line emission. The OH 1665 MHz line emission is detected at about 6 $σ$ level in the OH1 region by combining two epoch EVN observations. We found that the comp D1 shows the brightest CO, HCO+ line emission, as well as multi-band radio continuum emission. The environment around D1 shows no clear velocity structure associated with circular motions, making it different from most other OHMs in the literature, which might have been caused by an effect during the merger stage. Meanwhile, we found that the CO emission shows three velocity structures around D1, including the central broad FWHM region, the double peak region where the CO line profile shows two separated peaks, and the region of the high-velocity clouds where the CO line peaks at a high velocity ($\sim$ 11000 \kms). \HI in absorption also show high-velocity clouds around the D1 region, which might be due to inflows caused by the merging of two or more galaxy components. Based on the high-resolution K-band VLA and L-band VLBA observations of the radio continuum emission, we derived the brightness temperature in the range $10^{5}$ K to $10^{6}$ K, which is consistent with other starburst dominant OHM sources in the literature. The multi-band VLA observations show that the radio continuum emission of comp D might also have contributions from free-free emission, besides synchrotron emission. As a concenquence, these results support a starburst origin for the OHMs, without the presence of an AGN.

astro-ph.GA↗

arXiv2022

Un-Mix: Rethinking Image Mixtures for Unsupervised Visual Representation Learning

The recently advanced unsupervised learning approaches use the siamese-like framework to compare two "views" from the same image for learning representations. Making the two views distinctive is a core to guarantee that unsupervised methods can learn meaningful information. However, such frameworks are sometimes fragile on overfitting if the augmentations used for generating two views are not strong enough, causing the over-confident issue on the training data. This drawback hinders the model from learning subtle variance and fine-grained information. To address this, in this work we aim to involve the distance concept on label space in the unsupervised learning and let the model be aware of the soft degree of similarity between positive or negative pairs through mixing the input data space, to further work collaboratively for the input and loss spaces. Despite its conceptual simplicity, we show empirically that with the solution -- Unsupervised image mixtures (Un-Mix), we can learn subtler, more robust and generalized representations from the transformed input and corresponding new label space. Extensive experiments are conducted on CIFAR-10, CIFAR-100, STL-10, Tiny ImageNet and standard ImageNet with popular unsupervised methods SimCLR, BYOL, MoCo V1&V2, SwAV, etc. Our proposed image mixture and label assignment strategy can obtain consistent improvement by 1~3% following exactly the same hyperparameters and training procedures of the base methods. Code is publicly available at https://github.com/szq0214/Un-Mix.

cs.CV↗

arXiv2022

CHANG-ES. XXIV. First Detection of A Radio Nuclear Ring and Potential LLAGN in NGC 5792

We report the discoveries of a nuclear ring of diameter 10$\arcsec$ ($\sim$1.5 kpc) and a potential low luminosity active galactic nucleus (LLAGN) in the radio continuum emission map of the edge-on barred spiral galaxy NGC~5792. These discoveries are based on the Continuum Halos in Nearby Galaxies - an Expanded Very Large Array (VLA) Survey, as well as subsequent VLA observations of sub-arcsecond resolution. Using a mixture of H$α$ and 24 $μ$m calibration, we disentangle the thermal and non-thermal radio emission of the nuclear region, and derive a star formation rate (SFR) of $\sim 0.4~M_{\sun}$ yr$^{-1}$. We find that the nuclear ring is dominated by non-thermal synchrotron emission. The synchrotron-based SFR is about three times of the mixture-based SFR. This result indicates that the nuclear ring underwent more intense star-forming activity in the past, and now its star formation is in the low state. The sub-arcsecond VLA images resolve six individual knots on the nuclear ring. The equipartition magnetic field strength $B_{\rm eq}$ of the knots varies from 77 to 88 $μ$G. The radio ring surrounds a point-like faint radio core of $S_{\rm 6GHz}=(16\pm4)$ $μ$Jy with polarized lobes at the center of NGC~5792, which suggests an LLAGN with an Eddington ratio $\sim10^{-5}$. This radio nuclear ring is reminiscent of the Central Molecular Zone (CMZ) of the Galaxy. Both of them consist of a nuclear ring and LLAGN.

astro-ph.GA↗

arXiv2021

Structural and spectral properties of Galactic plane variable radio sources

In the time domain, the radio sky in particular along the Galactic plane direction may vary significantly because of various energetic activities associated with stars, stellar and supermassive black holes. Using multi-epoch Very Large Array surveys of the Galactic plane at 5.0 GHz, Becker et al. (2010) presented a catalogue of 39 variable radio sources in the flux density range 1-70 mJy. To probe their radio structures and spectra, we observed 17 sources with the very-long-baseline interferometric (VLBI) imaging technique and collected additional multi-frequency data from the literature. We detected all of the sources at 5 GHz with the Westerbork Synthesis Radio Telescope, but only G23.6644-0.0372 with the European VLBI Network (EVN). Together with its decadal variability and multi-frequency radio spectrum, we interpret it as an extragalactic peaked-spectrum source with a size of <~10 pc. The remaining sources were resolved out by the long baselines of the EVN because of either strong scatter broadening at the Galactic latitude <1 deg or intrinsically very extended structures on centi-arcsec scales. According to their spectral and structural properties, we find that the sample has a diverse nature. We notice two young H II regions and spot a radio star and a candidate planetary nebula. The rest of the sources are very likely associated with radio active galactic nuclei (AGN). Two of them also displays arcsec-scale faint jet activity. The sample study indicates that AGN are commonplace even among variable radio sources in the Galactic plane.

astro-ph.HE↗

arXiv2021

The intrinsic structure of Sagittarius A* at 1.3 cm and 7 mm

Sagittarius A* (Sgr A*), the Galactic Center supermassive black hole (SMBH), is one of the best targets to resolve the innermost region of SMBH with very long baseline interferometry (VLBI). In this study, we have carried out observations toward Sgr A* at 1.349 cm (22.223 GHz) and 6.950 mm (43.135 GHz) with the East Asian VLBI Network, as a part of the multi-wavelength campaign of the Event Horizon Telescope (EHT) in 2017 April. To mitigate scattering effects, the physically motivated scattering kernel model from Psaltis et al. (2018) and the scattering parameters from Johnson et al. (2018) have been applied. As a result, a single, symmetric Gaussian model well describes the intrinsic structure of Sgr A* at both wavelengths. From closure amplitudes, the major-axis sizes are ~704$\pm$102 $μ$as (axial ratio $\sim$1.19$^{+0.24}_{-0.19}$) and $\sim$300$\pm$25 $μ$as (axial ratio $\sim$1.28$\pm$0.2) at 1.349 cm and 6.95 mm respectively. Together with a quasi-simultaneous observation at 3.5 mm (86 GHz) by Issaoun et al. (2019), we show that the intrinsic size scales with observing wavelength as a power-law, with an index $\sim$1.2$\pm$0.2. Our results also provide estimates of the size and compact flux density at 1.3 mm, which can be incorporated into the analysis of the EHT observations. In terms of the origin of radio emission, we have compared the intrinsic structures with the accretion flow scenario, especially the radiatively inefficient accretion flow based on the Keplerian shell model. With this, we show that a nonthermal electron population is necessary to reproduce the source sizes.

astro-ph.HE↗

arXiv2021

A Fast Knowledge Distillation Framework for Visual Recognition

While Knowledge Distillation (KD) has been recognized as a useful tool in many visual tasks, such as supervised classification and self-supervised representation learning, the main drawback of a vanilla KD framework is its mechanism, which consumes the majority of the computational overhead on forwarding through the giant teacher networks, making the entire learning procedure inefficient and costly. ReLabel, a recently proposed solution, suggests creating a label map for the entire image. During training, it receives the cropped region-level label by RoI aligning on a pre-generated entire label map, allowing for efficient supervision generation without having to pass through the teachers many times. However, as the KD teachers are from conventional multi-crop training, there are various mismatches between the global label-map and region-level label in this technique, resulting in performance deterioration. In this study, we present a Fast Knowledge Distillation (FKD) framework that replicates the distillation training phase and generates soft labels using the multi-crop KD approach, while training faster than ReLabel since no post-processes such as RoI align and softmax operations are used. When conducting multi-crop in the same image for data loading, our FKD is even more efficient than the traditional image classification framework. On ImageNet-1K, we obtain 79.8% with ResNet-50, outperforming ReLabel by ~1.0% while being faster. On the self-supervised learning task, we also show that FKD has an efficiency advantage. Our project page: http://zhiqiangshen.com/projects/FKD/index.html, source code and models are available at: https://github.com/szq0214/FKD.

cs.CV↗

arXiv2021

Millimeter-VLBI observations of low-luminosity active galactic nuclei with source-frequency phase-referencing

We report millimeter-VLBI results of low-luminosity active galactic nuclei (M 84 and M 87) up to 88 GHz with source-frequency phase-referencing observations. We detected the weak VLBI core and obtained the first image of M 84 at 88 GHz. The derived brightness temperature of M 84 core was about 7.2$\times$10$^9$ K, which could serve as a lower limit as the core down to 30 Schwarzschild radii was still un-resolved in our 88 GHz observations. We successfully determined the core-shifts of M 87 at 22-44 GHz and 44-88 GHz through source-frequency phase-referencing technique. The jet apex of M 87 could be deduced at about 46 $μ$as upstream of the 43 GHz core from core-shift measurements. The estimated magnetic field strength of the 88 GHz core of M 87 is 4.8$\pm$2.4 G, which is at the same magnitude of 1-30 G near the event horizon probed by the Event Horizon Telescope.

astro-ph.GA↗

arXiv2021

Multi-modal Self-supervised Pre-training for Regulatory Genome Across Cell Types

In the genome biology research, regulatory genome modeling is an important topic for many regulatory downstream tasks, such as promoter classification, transaction factor binding sites prediction. The core problem is to model how regulatory elements interact with each other and its variability across different cell types. However, current deep learning methods often focus on modeling genome sequences of a fixed set of cell types and do not account for the interaction between multiple regulatory elements, making them only perform well on the cell types in the training set and lack the generalizability required in biological applications. In this work, we propose a simple yet effective approach for pre-training genome data in a multi-modal and self-supervised manner, which we call GeneBERT. Specifically, we simultaneously take the 1d sequence of genome data and a 2d matrix of (transcription factors x regions) as the input, where three pre-training tasks are proposed to improve the robustness and generalizability of our model. We pre-train our model on the ATAC-seq dataset with 17 million genome sequences. We evaluate our GeneBERT on regulatory downstream tasks across different cell types, including promoter classification, transaction factor binding sites prediction, disease risk estimation, and splicing sites prediction. Extensive experiments demonstrate the effectiveness of multi-modal and self-supervised pre-training for large-scale regulatory genomics data.

q-bio.GN↗

arXiv2021

Joint Multi-Dimension Pruning via Numerical Gradient Update

We present joint multi-dimension pruning (abbreviated as JointPruning), an effective method of pruning a network on three crucial aspects: spatial, depth and channel simultaneously. To tackle these three naturally different dimensions, we proposed a general framework by defining pruning as seeking the best pruning vector (i.e., the numerical value of layer-wise channel number, spacial size, depth) and construct a unique mapping from the pruning vector to the pruned network structures. Then we optimize the pruning vector with gradient update and model joint pruning as a numerical gradient optimization process. To overcome the challenge that there is no explicit function between the loss and the pruning vectors, we proposed self-adapted stochastic gradient estimation to construct a gradient path through network loss to pruning vectors and enable efficient gradient update. We show that the joint strategy discovers a better status than previous studies that focused on individual dimensions solely, as our method is optimized collaboratively across the three dimensions in a single end-to-end training and it is more efficient than the previous exhaustive methods. Extensive experiments on large-scale ImageNet dataset across a variety of network architectures MobileNet V1&V2&V3 and ResNet demonstrate the effectiveness of our proposed method. For instance, we achieve significant margins of 2.5% and 2.6% improvement over the state-of-the-art approach on the already compact MobileNet V1&V2 under an extremely large compression ratio.

cs.CV↗