Searcharxiv⌕ Search

arXiv subjects

Ke Zhang

Publications and source records attributed to Ke Zhang.

At least 217 records · Page 12Linked to original sources
arXiv2023

ReCLIP: Refine Contrastive Language Image Pre-Training with Source Free Domain Adaptation

Large-scale Pre-Training Vision-Language Model such as CLIP has demonstrated outstanding performance in zero-shot classification, e.g. achieving 76.3% top-1 accuracy on ImageNet without seeing any example, which leads to potential benefits to many tasks that have no labeled data. However, while applying CLIP to a downstream target domain, the presence of visual and text domain gaps and cross-modality misalignment can greatly impact the model performance. To address such challenges, we propose ReCLIP, the first source-free domain adaptation method for vision-language models, which does not require any source data or target labeled data. ReCLIP first learns a projection space to mitigate the misaligned visual-text embeddings and learns pseudo labels, and then deploys cross-modality self-training with the pseudo labels, to update visual and text encoders, refine labels and reduce domain gaps and misalignments iteratively. With extensive experiments, we demonstrate ReCLIP reduces the average error rate of CLIP from 30.17% to 25.06% on 22 image classification benchmarks. Code available at https://github.com/michiganleon/ReCLIP_WACV.

cs.CV↗
arXiv2023

Water-Rich Disks around Late M-stars Unveiled: Exploring the Remarkable Case of Sz114

We present an analysis of the JDISC JWST/MIRI-MRS spectrum of Sz~114, an accreting M5 star surrounded by a large dust disk with a shallow gap at $\sim 39$ au. The spectrum is molecular-rich: we report the detection of water, CO, CO$_2$, HCN, C$_2$H$_2$, and H$_2$. The only identified atomic/ionic transition is from [NeII] at 12.81 micron. A distinct feature of this spectrum is the forest of water lines with the 17.22 micron emission surpassing that of most mid-to-late M-star disks by an order of magnitude in flux and aligning instead with disks of earlier-type stars. Moreover, flux ratios of C$_2$H$_2$/H$_2$O and HCN/H$_2$O in Sz~114 also resemble those of earlier-type disks, with a slightly elevated CO$_2$/H$_2$O ratio. While accretional heating can boost all infrared lines, the unusual properties of Sz~114 could be explained by the young age of the source, its formation under unusual initial conditions (a large massive disk), and the presence of dust substructures. The latter delays the inward drift of icy pebbles and help preserve a lower C/O ratio over an extended period. In contrast, mid-to-late M-star disks--which are typically faint, small in size, and likely lack significant substructures--may have more quickly depleted the outer icy reservoir and already evolved out of a water-rich inner disk phase. Our findings underscore the unexpected diversity within mid-infrared spectra of mid-to-late M-star disks, highlighting the need to expand the observational sample for a comprehensive understanding of their variations and thoroughly test pebble drift and planet formation models.

astro-ph.EP↗
arXiv2023

Multi-task Paired Masking with Alignment Modeling for Medical Vision-Language Pre-training

In recent years, the growing demand for medical imaging diagnosis has placed a significant burden on radiologists. As a solution, Medical Vision-Language Pre-training (Med-VLP) methods have been proposed to learn universal representations from medical images and reports, benefiting downstream tasks without requiring fine-grained annotations. However, existing methods have overlooked the importance of cross-modal alignment in joint image-text reconstruction, resulting in insufficient cross-modal interaction. To address this limitation, we propose a unified Med-VLP framework based on Multi-task Paired Masking with Alignment (MPMA) to integrate the cross-modal alignment task into the joint image-text reconstruction framework to achieve more comprehensive cross-modal interaction, while a Global and Local Alignment (GLA) module is designed to assist self-supervised paradigm in obtaining semantic representations with rich domain knowledge. Furthermore, we introduce a Memory-Augmented Cross-Modal Fusion (MA-CMF) module to fully integrate visual information to assist report reconstruction and fuse the multi-modal representations adequately. Experimental results demonstrate that the proposed unified approach outperforms previous methods in all downstream tasks, including uni-modal, cross-modal, and multi-modal tasks.

cs.CV↗
arXiv2023

JWST reveals excess cool water near the snowline in compact disks, consistent with pebble drift

Previous analyses of mid-infrared water spectra from young protoplanetary disks observed with the Spitzer-IRS found an anti-correlation between water luminosity and the millimeter dust disk radius observed with ALMA. This trend was suggested to be evidence for a fundamental process of inner disk water enrichment, used to explain properties of the Solar System 40 years ago, in which icy pebbles drift inward from the outer disk and sublimate after crossing the snowline. Previous analyses of IRS water spectra, however, were uncertain due to the low spectral resolution that blended lines together. We present new JWST-MIRI spectra of four disks, two compact and two large with multiple radial gaps, selected to test the scenario that water vapor inside the snowline is regulated by pebble drift. The higher spectral resolving power of MIRI-MRS now yields water spectra that separate individual lines, tracing upper level energies from 900 K to 10,000 K. These spectra clearly reveal excess emission in the low-energy lines in compact disks, compared to the large disks, demonstrating an enhanced cool component with $T \approx$ 170-400 K and equivalent emitting radius $R_{\rm{eq}}\approx$ 1-10 au. We interpret the cool water emission as ice sublimation and vapor diffusion near the snowline, suggesting that there is indeed a higher inwards mass flux of icy pebbles in compact disks. Observation of this process opens up multiple exciting prospects to study planet formation chemistry in inner disks with JWST.

astro-ph.EP↗
arXiv2023

ImGeoNet: Image-induced Geometry-aware Voxel Representation for Multi-view 3D Object Detection

We propose ImGeoNet, a multi-view image-based 3D object detection framework that models a 3D space by an image-induced geometry-aware voxel representation. Unlike previous methods which aggregate 2D features into 3D voxels without considering geometry, ImGeoNet learns to induce geometry from multi-view images to alleviate the confusion arising from voxels of free space, and during the inference phase, only images from multiple views are required. Besides, a powerful pre-trained 2D feature extractor can be leveraged by our representation, leading to a more robust performance. To evaluate the effectiveness of ImGeoNet, we conduct quantitative and qualitative experiments on three indoor datasets, namely ARKitScenes, ScanNetV2, and ScanNet200. The results demonstrate that ImGeoNet outperforms the current state-of-the-art multi-view image-based method, ImVoxelNet, on all three datasets in terms of detection accuracy. In addition, ImGeoNet shows great data efficiency by achieving results comparable to ImVoxelNet with 100 views while utilizing only 40 views. Furthermore, our studies indicate that our proposed image-induced geometry-aware representation can enable image-based methods to attain superior detection accuracy than the seminal point cloud-based method, VoteNet, in two practical scenarios: (1) scenarios where point clouds are sparse and noisy, such as in ARKitScenes, and (2) scenarios involve diverse object classes, particularly classes of small objects, as in the case in ScanNet200.

cs.CV↗
arXiv2023

How large is a disk -- what do protoplanetary disk gas sizes really mean?

It remains unclear what mechanism is driving the evolution of protoplanetary disks. Direct detection of the main candidates, either turbulence driven by magnetorotational instability or magnetohydrodynamical disk winds, has proven difficult, leaving the time evolution of the disk size as one of the most promising observables able to differentiate between these two mechanisms. But to do so successfully, we need to understand what the observed gas disk size actually traces. We studied the relation between $R_{\rm CO,\ 90\%}$, the radius that encloses 90% of the $^{12}$CO flux, and $R_c$, the radius that encodes the physical disk size, in order to provide simple prescriptions for conversions between these two sizes. For an extensive grid of thermochemical models we calculate $R_{\rm CO,\ 90\%}$ from synthetic observations and relate properties measured at this radius, such as the gas column density, to bulk disk properties, such as $R_c$ and the disk mass $M_{\rm disk}$. We found an empirical correlation between the gas column density at $R_{\rm CO,\ 90\%}$ and disk mass: $N_{\rm gas}(R_{\rm CO,\ 90\%}) \approx 3.73\times10^{21}(M_{\rm disk}/\mathrm{M}_{\odot})^{0.34}\ \mathrm{cm}^{-2}$. Using this correlation we derive an analytical prescription of $R_{\rm CO,\ 90\%}$ that only depends on $R_c$ and $M_{\rm disk}$. We derive $R_c$ for disks in Lupus, Upper Sco, Taurus and DSHARP, finding that disks in the older Upper Sco region are significantly smaller ($\langle R_c \rangle$ = 4.8 au) than disks in the younger Lupus and Taurus regions ($\langle R_c \rangle$ = 19.8 and 20.9 au, respectively). This temporal decrease in $R_c$ goes against predictions of both viscous and wind-driven evolution, but could be a sign of significant external photoevaporation having truncated disks in Upper Sco.

astro-ph.EP↗
arXiv2023

Spatio-Temporal Classification of Lung Ventilation Patterns using 3D EIT Images: A General Approach for Individualized Lung Function Evaluation

The Pulmonary Function Test (PFT) is an widely utilized and rigorous classification test for lung function evaluation, serving as a comprehensive tool for lung diagnosis. Meanwhile, Electrical Impedance Tomography (EIT) is a rapidly advancing clinical technique that visualizes conductivity distribution induced by ventilation. EIT provides additional spatial and temporal information on lung ventilation beyond traditional PFT. However, relying solely on conventional isolated interpretations of PFT results and EIT images overlooks the continuous dynamic aspects of lung ventilation. This study aims to classify lung ventilation patterns by extracting spatial and temporal features from the 3D EIT image series. The study uses a Variational Autoencoder network with a MultiRes block to compress the spatial distribution in a 3D image into a one-dimensional vector. These vectors are then concatenated to create a feature map for the exhibition of temporal features. A simple convolutional neural network is used for classification. Data collected from 137 subjects were finally used for training. The model is validated by ten-fold and leave-one-out cross-validation first. The accuracy and sensitivity of normal ventilation mode are 0.95 and 1.00, and the f1-score is 0.94. Furthermore, we check the reliability and feasibility of the proposed pipeline by testing it on newly recruited nine subjects. Our results show that the pipeline correctly predicts the ventilation mode of 8 out of 9 subjects. The study demonstrates the potential of using image series for lung ventilation mode classification, providing a feasible method for patient prescreening and presenting an alternative form of PFT.

eess.IV↗
arXiv2023

BUMP: A Benchmark of Unfaithful Minimal Pairs for Meta-Evaluation of Faithfulness Metrics

The proliferation of automatic faithfulness metrics for summarization has produced a need for benchmarks to evaluate them. While existing benchmarks measure the correlation with human judgements of faithfulness on model-generated summaries, they are insufficient for diagnosing whether metrics are: 1) consistent, i.e., indicate lower faithfulness as errors are introduced into a summary, 2) effective on human-written texts, and 3) sensitive to different error types (as summaries can contain multiple errors). To address these needs, we present a benchmark of unfaithful minimal pairs (BUMP), a dataset of 889 human-written, minimally different summary pairs, where a single error is introduced to a summary from the CNN/DailyMail dataset to produce an unfaithful summary. We find BUMP complements existing benchmarks in a number of ways: 1) the summaries in BUMP are harder to discriminate and less probable under SOTA summarization models, 2) unlike non-pair-based datasets, BUMP can be used to measure the consistency of metrics, and reveals that the most discriminative metrics tend not to be the most consistent, and 3) unlike datasets containing generated summaries with multiple errors, BUMP enables the measurement of metrics' performance on individual error types.

cs.CL↗
arXiv2023

Substructures in Compact Disks of the Taurus Star-forming Region

Observations of substructure in protoplanetary disks have largely been limited to the brightest and largest disks, excluding the abundant population of compact disks which are likely sites of planet formation. Here, we reanalyze ~0.1'', 1.33 mm ALMA continuum observations of 12 compact protoplanetary disks in the Taurus star-forming region. By fitting visibilities directly, we identify substructures in 6 of the 12 compact disks. We then compare the substructures identified in the full Taurus sample of 24 disks in single star systems and the ALMA DSHARP survey, differentiating between compact (R_eff,90% < 50 au) and extended (R_eff,90% > 50 au) disk sources. We find that substructures are detected at nearly all radii in both small and large disks. Tentatively, we find fewer wide gaps in intermediate-sized disks with R_eff,90% between 30 and 90 au. We perform a series of planet-disk interaction simulations to constrain the sensitivity of our visibility-fitting approach. Under an assumption of planet-disk interaction, we use the gap widths and common disk parameters to calculate potential planet masses within the Taurus sample. We find that the young planet occurrence rate peaks near Neptune masses, similar to the DSHARP sample. For 0.01 $M_J/M_\odot$ $\lesssim$ $M_p/M_*$ $\lesssim$ 0.1 $M_J/M_\odot$, the rate is 17.4$\pm$8.3%; for 0.1 $M_J/M_\odot$ $\lesssim$ $M_p/M_*$ $\lesssim$ 1 $M_J/M_\odot$, it is 27.8$\pm$8.3%. Both of them are consistent with microlensing surveys. For gas giants more massive than 5 $M_J$, the occurrence rate is 4.2$\pm$4.2%, consistent with direct imaging surveys.

astro-ph.EP↗
arXiv2023

Predict NAS Multi-Task by Stacking Ensemble Models using GP-NAS

Accurately predicting the performance of architecture with small sample training is an important but not easy task. How to analysis and train dataset to overcome overfitting is the core problem we should deal with. Meanwhile if there is the mult-task problem, we should also think about if we can take advantage of their correlation and estimate as fast as we can. In this track, Super Network builds a search space based on ViT-Base. The search space contain depth, num-heads, mpl-ratio and embed-dim. What we done firstly are pre-processing the data based on our understanding of this problem which can reduce complexity of problem and probability of over fitting. Then we tried different kind of models and different way to combine them. Finally we choose stacking ensemble models using GP-NAS with cross validation. Our stacking model ranked 1st in CVPR 2022 Track 2 Challenge.

cs.LG↗
arXiv2023

Adjust factor with volatility model using MAXFLAT low-pass filter and construct portfolio in China A share market

In the field of quantitative finance, volatility models, such as ARCH, GARCH, FIGARCH, SV, EWMA, play the key role in risk and portfolio management. Meanwhile, factor investing is more and more famous since mid of 20 century. CAPM, Fama French three factor model, Fama French five-factor model, MSCI Barra factor model are mentioned and developed during this period. In this paper, we will show why we need adjust group of factors by our MAXFLAT low-pass volatility model. All of our experiments are under China's CSI 300 and CSI 500 universe which represent China's large cap stocks and mid-small cap stocks. Our result shows adjust factors by MAXFLAT volatility model have better performance in both large cap and small cap universe than original factors or other risk adjust factors in China A share. Also the portfolio constructed by MAXFLAT risk adjust factors have continuous excess return and lower beta compare with benchmark index.

q-fin.RM↗
arXiv2023

Construct sparse portfolio with mutual fund's favourite stocks in China A share market

Unlike developed market, some emerging markets are dominated by retail and unprofessional trading. China A share market is a good and fitting example in last 20 years. Meanwhile, lots of research show professional investor in China A share market continuously generate excess return compare with total market index. Specifically, this excess return mostly come from stock selectivity ability instead of market timing. However for some reason such as fund capacity limit, fund manager change or market regional switch, it is very hard to find a fund could continuously beat market. Therefore, in order to get excess return from mutual fund industry, we use quantitative way to build the sparse portfolio that take advantage of favorite stocks by mutual fund in China A market. Firstly we do the analysis about favourite stocks by mutual fund and compare the different method to construct our portfolio. Then we build a sparse stock portfolio with constraint on both individual stock and industry exposure using portfolio optimizer to closely track the partial equity funds index 930950.CSI with median 0.985 correlation. This problem is much more difficult than tracking full information index or traditional ETF as higher turnover of mutual fund, just first 10 holding of mutual fund available and fund report updated quarterly with 15 days delay. Finally we build another low risk and balanced sparse portfolio that consistently outperform benchmark 930950.CSI.

q-fin.PM↗
arXiv2023

Evaluating the Effectiveness of 2D and 3D Features for Predicting Tumor Response to Chemotherapy

2D and 3D tumor features are widely used in a variety of medical image analysis tasks. However, for chemotherapy response prediction, the effectiveness between different kinds of 2D and 3D features are not comprehensively assessed, especially in ovarian cancer-related applications. This investigation aims to accomplish such a comprehensive evaluation. For this purpose, CT images were collected retrospectively from 188 advanced-stage ovarian cancer patients. All the metastatic tumors that occurred in each patient were segmented and then processed by a set of six filters. Next, three categories of features, namely geometric, density, and texture features, were calculated from both the filtered results and the original segmented tumors, generating a total of 1595 and 1403 features for the 3D and 2D tumors, respectively. In addition to the conventional single-slice 2D and full-volume 3D tumor features, we also computed the incomplete-3D tumor features, which were achieved by sequentially adding one individual CT slice and calculating the corresponding features. Support vector machine (SVM) based prediction models were developed and optimized for each feature set. 5-fold cross-validation was used to assess the performance of each individual model. The results show that the 2D feature-based model achieved an AUC (area under the ROC curve [receiver operating characteristic]) of 0.84+-0.02. When adding more slices, the AUC first increased to reach the maximum and then gradually decreased to 0.86+-0.02. The maximum AUC was yielded when adding two adjacent slices, with a value of 0.91+-0.01. This initial result provides meaningful information for optimizing machine learning-based decision-making support tools in the future.

eess.IV↗
arXiv2023

Design and resonator-assisted characterization of high performance lithium niobate waveguide crossings

Waveguide crossings are elementary passive components for signal routing in photonic integrated circuits. Here, we design and characterize two multimode interferometer (MMI)-based waveguide crossings to serve the various routing directions in the anisotropic x-cut thin-film lithium niobate (TFLN) platform. To address the large measurement uncertainties in traditional cut-back characterization methods, we propose and demonstrate a resonator-assisted approach that dramatically reduces the uncertainty of insertion loss measurement(< 0.021 dB) and the lower bound of crosstalk measurement (-60 dB) using only two devices. Based on this approach, we demonstrate and verify TFLN waveguide crossings with insertion losses of < 0.070 dB and crosstalk of < -50 dB along all three routing directions at 1550 nm. The low-loss and low-crosstalk waveguide crossings in this work, together with the simple and efficient characterization strategy, could provide important layout design flexibility for future large-scale classical and quantum TFLN photonic circuits.

physics.optics↗
arXiv2023

A gap-sharing planet pair shaping the crescent in HD 163296: a disk sculpted by a resonant chain

ALMA observations of the disk around HD 163296 have resolved a crescent-shape substructure at around 55 au, inside and off-center from a gap in the dust that extends from 38 au to 62 au. In this work we propose that both the crescent and the dust rings are caused by a compact pair (period ratio $\simeq 4:3$) of sub-Saturn-mass planets inside the gap, with the crescent corresponding to dust trapped at the $L_5$ Lagrange point of the outer planet. This interpretation also reproduces well the gap in the gas recently measured from the CO observations, which is shallower than what is expected in a model where the gap is carved by a single planet. Building on previous works arguing for outer planets at $\approx 86$ and $\approx 137$ au, we provide with a global model of the disk that best reproduces the data and show that all four planets may fall into a long resonant chain, with the outer three planets in a 1:2:4 Laplace resonance. We show that this configuration is not only an expected outcome from disk-planet interaction in this system, but it can also help constraining the radial and angular position of the planet candidates using three-body resonances.

astro-ph.EP↗
arXiv2023

ZScribbleSeg: Zen and the Art of Scribble Supervised Medical Image Segmentation

Curating a large scale fully-annotated dataset can be both labour-intensive and expertise-demanding, especially for medical images. To alleviate this problem, we propose to utilize solely scribble annotations for weakly supervised segmentation. Existing solutions mainly leverage selective losses computed solely on annotated areas and generate pseudo gold standard segmentation by propagating labels to adjacent areas. However, these methods could suffer from the inaccurate and sometimes unrealistic pseudo segmentation due to the insufficient supervision and incomplete shape features. Different from previous efforts, we first investigate the principle of ''good scribble annotations'', which leads to efficient scribble forms via supervision maximization and randomness simulation. Furthermore, we introduce regularization terms to encode the spatial relationship and shape prior, where a new formulation is developed to estimate the mixture ratios of label classes. These ratios are critical in identifying the unlabeled pixels for each class and correcting erroneous predictions, thus the accurate estimation lays the foundation for the incorporation of spatial prior. Finally, we integrate the efficient scribble supervision with the prior into a unified framework, denoted as ZScribbleSeg, and apply the method to multiple scenarios. Leveraging only scribble annotations, ZScribbleSeg set new state-of-the-arts on four segmentation tasks using ACDC, MSCMRseg, MyoPS and PPSS datasets.

cs.CV↗
arXiv2023

CameraPose: Weakly-Supervised Monocular 3D Human Pose Estimation by Leveraging In-the-wild 2D Annotations

To improve the generalization of 3D human pose estimators, many existing deep learning based models focus on adding different augmentations to training poses. However, data augmentation techniques are limited to the "seen" pose combinations and hard to infer poses with rare "unseen" joint positions. To address this problem, we present CameraPose, a weakly-supervised framework for 3D human pose estimation from a single image, which can not only be applied on 2D-3D pose pairs but also on 2D alone annotations. By adding a camera parameter branch, any in-the-wild 2D annotations can be fed into our pipeline to boost the training diversity and the 3D poses can be implicitly learned by reprojecting back to 2D. Moreover, CameraPose introduces a refinement network module with confidence-guided loss to further improve the quality of noisy 2D keypoints extracted by 2D pose estimators. Experimental results demonstrate that the CameraPose brings in clear improvements on cross-scenario datasets. Notably, it outperforms the baseline method by 3mm on the most challenging dataset 3DPW. In addition, by combining our proposed refinement network module with existing 3D pose estimators, their performance can be improved in cross-scenario evaluation.

cs.CV↗
arXiv2022

Multi-head Uncertainty Inference for Adversarial Attack Detection

Deep neural networks (DNNs) are sensitive and susceptible to tiny perturbation by adversarial attacks which causes erroneous predictions. Various methods, including adversarial defense and uncertainty inference (UI), have been developed in recent years to overcome the adversarial attacks. In this paper, we propose a multi-head uncertainty inference (MH-UI) framework for detecting adversarial attack examples. We adopt a multi-head architecture with multiple prediction heads (i.e., classifiers) to obtain predictions from different depths in the DNNs and introduce shallow information for the UI. Using independent heads at different depths, the normalized predictions are assumed to follow the same Dirichlet distribution, and we estimate distribution parameter of it by moment matching. Cognitive uncertainty brought by the adversarial attacks will be reflected and amplified on the distribution. Experimental results show that the proposed MH-UI framework can outperform all the referred UI methods in the adversarial attack detection task with different settings.

cs.LG↗