arXiv subjects

Jiaqi Li

Publications and source records attributed to Jiaqi Li.

At least 199 records · Page 11Linked to original sources

arXiv2023

When Source-Free Domain Adaptation Meets Learning with Noisy Labels

Recent state-of-the-art source-free domain adaptation (SFDA) methods have focused on learning meaningful cluster structures in the feature space, which have succeeded in adapting the knowledge from source domain to unlabeled target domain without accessing the private source data. However, existing methods rely on the pseudo-labels generated by source models that can be noisy due to domain shift. In this paper, we study SFDA from the perspective of learning with label noise (LLN). Unlike the label noise in the conventional LLN scenario, we prove that the label noise in SFDA follows a different distribution assumption. We also prove that such a difference makes existing LLN methods that rely on their distribution assumptions unable to address the label noise in SFDA. Empirical evidence suggests that only marginal improvements are achieved when applying the existing LLN methods to solve the SFDA problem. On the other hand, although there exists a fundamental difference between the label noise in the two scenarios, we demonstrate theoretically that the early-time training phenomenon (ETP), which has been previously observed in conventional label noise settings, can also be observed in the SFDA problem. Extensive experiments demonstrate significant improvements to existing SFDA algorithms by leveraging ETP to address the label noise in SFDA.

cs.LG↗

arXiv2023

Towards Accurate Acne Detection via Decoupled Sequential Detection Head

Accurate acne detection plays a crucial role in acquiring precise diagnosis and conducting proper therapy. However, the ambiguous boundaries and arbitrary dimensions of acne lesions severely limit the performance of existing methods. In this paper, we address these challenges via a novel Decoupled Sequential Detection Head (DSDH), which can be easily adopted by mainstream two-stage detectors. DSDH brings two simple but effective improvements to acne detection. Firstly, the offset and scaling tasks are explicitly introduced, and their incompatibility is settled by our task-decouple mechanism, which improves the capability of predicting the location and size of acne lesions. Second, we propose the task-sequence mechanism, and execute offset and scaling sequentially to gain a more comprehensive insight into the dimensions of acne lesions. In addition, we build a high-quality acne detection dataset named ACNE-DET to verify the effectiveness of DSDH. Experiments on ACNE-DET and the public benchmark ACNE04 show that our method outperforms the state-of-the-art methods by significant margins. Our code and dataset are publicly available at (temporarily anonymous).

cs.CV↗

arXiv2022

On Learning Fairness and Accuracy on Multiple Subgroups

We propose an analysis in fair learning that preserves the utility of the data while reducing prediction disparities under the criteria of group sufficiency. We focus on the scenario where the data contains multiple or even many subgroups, each with limited number of samples. As a result, we present a principled method for learning a fair predictor for all subgroups via formulating it as a bilevel objective. Specifically, the subgroup specific predictors are learned in the lower-level through a small amount of data and the fair predictor. In the upper-level, the fair predictor is updated to be close to all subgroup specific predictors. We further prove that such a bilevel objective can effectively control the group sufficiency and generalization error. We evaluate the proposed framework on real-world datasets. Empirical evidence suggests the consistently improved fair predictions, as well as the comparable accuracy to the baselines.

stat.ML↗

arXiv2022

MNER-QG: An End-to-End MRC framework for Multimodal Named Entity Recognition with Query Grounding

Multimodal named entity recognition (MNER) is a critical step in information extraction, which aims to detect entity spans and classify them to corresponding entity types given a sentence-image pair. Existing methods either (1) obtain named entities with coarse-grained visual clues from attention mechanisms, or (2) first detect fine-grained visual regions with toolkits and then recognize named entities. However, they suffer from improper alignment between entity types and visual regions or error propagation in the two-stage manner, which finally imports irrelevant visual information into texts. In this paper, we propose a novel end-to-end framework named MNER-QG that can simultaneously perform MRC-based multimodal named entity recognition and query grounding. Specifically, with the assistance of queries, MNER-QG can provide prior knowledge of entity types and visual regions, and further enhance representations of both texts and images. To conduct the query grounding task, we provide manual annotations and weak supervisions that are obtained via training a highly flexible visual grounding model with transfer learning. We conduct extensive experiments on two public MNER datasets, Twitter2015 and Twitter2017. Experimental results show that MNER-QG outperforms the current state-of-the-art models on the MNER task, and also improves the query grounding performance.

cs.CV↗

arXiv2022

Efficient Single-Image Depth Estimation on Mobile Devices, Mobile AI & AIM 2022 Challenge: Report

Various depth estimation models are now widely used on many mobile and IoT devices for image segmentation, bokeh effect rendering, object tracking and many other mobile tasks. Thus, it is very crucial to have efficient and accurate depth estimation models that can run fast on low-power mobile chipsets. In this Mobile AI challenge, the target was to develop deep learning-based single image depth estimation solutions that can show a real-time performance on IoT platforms and smartphones. For this, the participants used a large-scale RGB-to-depth dataset that was collected with the ZED stereo camera capable to generated depth maps for objects located at up to 50 meters. The runtime of all models was evaluated on the Raspberry Pi 4 platform, where the developed solutions were able to generate VGA resolution depth maps at up to 27 FPS while achieving high fidelity results. All models developed in the challenge are also compatible with any Android or Linux-based mobile devices, their detailed description is provided in this paper.

cs.CV↗

arXiv2022

Imaging-based 3D Particle Tracking System for Field Characterization of Particle Dynamics in Atmospheric Flows

A particle tracking velocimetry apparatus is presented that is capable of measuring three-dimensional particle trajectories across large volumes, of the order of several meters, during natural snowfall events. Field experiments, aimed at understanding snow settling kinematics in atmospheric flows, were conducted during the 2021/2022 winter season using this apparatus, from which we show preliminary results. An overview of the methodology, wherein we use a UAV-based calibration method, is provided, and analysis is conducted of a select dataset to demonstrate the capabilities of the system for studying inertial particle dynamics in atmospheric flows. A modular camera array is used, designed specifically for handling the challenges of field deployment during snowfall. This imaging system is calibrated using synchronized imaging of a UAV-carried target to enable measurements centered 10 m above the ground within approximately a 4 m x 4 m x 6 m volume. Using the measured Lagrangian particle tracks we present data concerning 3D trajectory curvature and acceleration statistics, as well as clustering behavior using Voronoi analysis. The limitations, as well as potential future developments, of such a system are discussed in the context of applications with other inertial particles.

physics.flu-dyn↗

arXiv2022

Snow particle analyzer for simultaneous measurements of snow density and morphology

The detailed characterization of snow particles is critical for understanding the snow settling behavior and modeling the ground snow accumulation for various applications such as prevention of avalanches and snowmelt-caused floods, etc. In this study, we present a snow particle analyzer for simultaneous measurements of various properties of fresh falling snow, including their concentration, size, shape, type, and density. The analyzer consists of a digital inline holography module for imaging falling snow particles in a sample volume of 88 cm3 and a high-precision scale to measure the weight of the same particles in a synchronized fashion. The holographic images are processed in real-time using a machine learning model and post-processing to determine snow particle concentration, size, shape, and type. Such information is used to obtain the estimated volume, which is subsequently correlated with the weight of snow particles to estimate their density. The performance of the analyzer is assessed using monodispersed spherical glass beads and irregular salt crystals with known density, which shows <5% density measurement errors. In addition, the analyzer was tested in a number of field deployments under different snow and wind conditions. The system is able to achieve measurements of various snow properties at single particle resolution and statistical robustness. The analyzer was also deployed for four hours of operation during a snow event with changing snow and wind conditions, demonstrating its ability for long-term and real-time monitoring of the time-varying snow properties in the field.

physics.ao-ph↗

arXiv2022

Automated Noncontact Trapping of Moving Micro-particle with Ultrasonic Phased Array System and Microscopic Vision

Noncontact particle manipulation (NPM) technology has significantly extended mankind's analysis capability into micro and nano scale, which in turn greatly promoted the development of material science and life science. Though NPM by means of electric, magnetic, and optical field has achieved great success, from the robotic perspective, it is still labor-intensive manipulation since professional human assistance is somehow mandatory in early preparation stage. Therefore, developing automated noncontact trapping of moving particles is worthwhile, particularly for applications where particle samples are rare, fragile or contact sensitive. Taking advantage of latest dynamic acoustic field modulating technology, and particularly by virtue of the great scalability of acoustic manipulation from micro-scale to sub-centimeter-scale, we propose an automated noncontact trapping of moving micro-particles with ultrasonic phased array system and microscopic vision in this paper. The main contribution of this work is for the first time, as far as we know, we achieved fully automated moving micro-particle trapping in acoustic NPM field by resorting to robotic approach. In short, the particle moving status is observed and predicted by binocular microscopic vision system, by referring to which the acoustic trapping zone is calculated and generated to capture and stably hold the particle. The problem of hand-eye relationship of noncontact robotic end-effector is also solved in this work. Experiments demonstrated the effectiveness of this work.

cs.RO↗

arXiv2022

Direct measurement of vorticity using tracer particles with internal markers

Current experiment techniques for vorticity measurement suffer from limited spatial and temporal resolution to resolve the small-scale eddy dynamics in turbulence. In this study, we develop a new method for direct vorticity measurement in fluid flows based on digital inline holography (DIH). The DIH system utilizes a collimated laser beam to illuminate the tracers with internal markers and a digital sensor to record the generated holograms. The tracers made of the polydimethylsiloxane (PDMS) prepolymer mixed with internal markers are fabricated using a standard microfluidic droplet generator. A rotation measurement algorithm is developed based on the 3D location reconstruction and tracking of the internal markers and is assessed through synthetic holograms to identify the optimal parameter settings and measurement range (e.g., rotation rate from 0.3 to 0.7 rad/frame under numerical aperture of imaging of 0.25). Our proposed method based on DIH is evaluated by a calibration experiment of single tracer rotation, which yields the same optimal measurement range. Using von Kármán swirling flow setup, we further demonstrate the capability of the approach to simultaneously measure the Lagrangian rotation and translation of multiple tracers. Our method can measure vorticity in a small region on the order of 100 $μ$m or less and can be potentially used to quantify the Kolmogorov-scale vorticity field in turbulent flows.

physics.flu-dyn↗

arXiv2022

CrossHuman: Learning Cross-Guidance from Multi-Frame Images for Human Reconstruction

We propose CrossHuman, a novel method that learns cross-guidance from parametric human model and multi-frame RGB images to achieve high-quality 3D human reconstruction. To recover geometry details and texture even in invisible regions, we design a reconstruction pipeline combined with tracking-based methods and tracking-free methods. Given a monocular RGB sequence, we track the parametric human model in the whole sequence, the points (voxels) corresponding to the target frame are warped to reference frames by the parametric body motion. Guided by the geometry priors of the parametric body and spatially aligned features from RGB sequence, the robust implicit surface is fused. Moreover, a multi-frame transformer (MFT) and a self-supervised warp refinement module are integrated to the framework to relax the requirements of parametric body and help to deal with very loose cloth. Compared with previous works, our CrossHuman enables high-fidelity geometry details and texture in both visible and invisible regions and improves the accuracy of the human reconstruction even under estimated inaccurate parametric human models. The experiments demonstrate that our method achieves state-of-the-art (SOTA) performance.

cs.CV↗

arXiv2022

Learning High-quality Proposals for Acne Detection

Acne detection is crucial for interpretative diagnosis and precise treatment of skin disease. The arbitrary boundary and small size of acne lesions lead to a significant number of poor-quality proposals in two-stage detection. In this paper, we propose a novel head structure for Region Proposal Network to improve the proposals' quality in two ways. At first, a Spatial Aware Double Head(SADH) structure is proposed to disentangle the representation learning for classification and localization from two different spatial perspectives. The proposed SADH ensures a steeper classification confidence gradient and suppresses the proposals having low intersection-over-union(IoU) with the matched ground truth. Then, we propose a Normalized Wasserstein Distance prediction branch to improve the correlation between the proposals' classification scores and IoUs. In addition, to facilitate further research on acne detection, we construct a new dataset named AcneSCU, with high-resolution imageries, precise annotations, and fine-grained lesion categories. Extensive experiments are conducted on both AcneSCU and the public dataset ACNE04, and the results demonstrate the proposed method could improve the proposals' quality, consistently outperforming state-of-the-art approaches. Code and the collected dataset are available in https://github.com/pingguokiller/acnedetection.

cs.CV↗

arXiv2022

Evolving Domain Generalization

Domain generalization aims to learn a predictive model from multiple different but related source tasks that can generalize well to a target task without the need of accessing any target data. Existing domain generalization methods ignore the relationship between tasks, implicitly assuming that all the tasks are sampled from a stationary environment. Therefore, they can fail when deployed in an evolving environment. To this end, we formulate and study the \emph{evolving domain generalization} (EDG) scenario, which exploits not only the source data but also their evolving pattern to generate a model for the unseen task. Our theoretical result reveals the benefits of modeling the relation between two consecutive tasks by learning a globally consistent directional mapping function. In practice, our analysis also suggests solving the DDG problem in a meta-learning manner, which leads to \emph{directional prototypical network}, the first method for the DDG problem. Empirical evaluation of both synthetic and real-world data sets validates the effectiveness of our approach.

cs.LG↗

arXiv2022

Fair Representation Learning through Implicit Path Alignment

We consider a fair representation learning perspective, where optimal predictors, on top of the data representation, are ensured to be invariant with respect to different sub-groups. Specifically, we formulate this intuition as a bi-level optimization, where the representation is learned in the outer-loop, and invariant optimal group predictors are updated in the inner-loop. Moreover, the proposed bi-level objective is demonstrated to fulfill the sufficiency rule, which is desirable in various practical scenarios but was not commonly studied in the fair learning. Besides, to avoid the high computational and memory cost of differentiating in the inner-loop of bi-level objective, we propose an implicit path alignment algorithm, which only relies on the solution of inner optimization and the implicit differentiation rather than the exact optimization path. We further analyze the error gap of the implicit approach and empirically validate the proposed method in both classification and regression settings. Experimental results show the consistently better trade-off in prediction performance and fairness measurement.

cs.LG↗

arXiv2022

Nonreciprocal waveguide-QED for spinning cavities with multiple coupling points

We investigate chiral emission and the single-photon scattering of spinning cavities coupled to a meandering waveguide at multiple coupling points. It is shown that nonreciprocal photon transmissions occur in the cavities-waveguide system, which stems from interference effects among different coupling points, and frequency shifts induced by the Sagnac effect. The nonlocal interference is akin to the mechanism in giant atoms. In the single-cavity setup, by optimizing the spinning velocity and number of coupling points, the chiral factor can approach 1, and the chiral direction can be freely switched. Moreover, destructive interference gives rise to the complete photon transmission in one direction over the whole optical frequency band, with no analogy in other quantum setups. In the multiple-cavity system, we also investigate the photon transport properties. The results indicate a directional information flow between different nodes. Our proposal provides a novel way to achieve quantum nonreciprocal devices, which can be applied in large-scale quantum chiral networks with optical waveguides.

quant-ph↗

arXiv2022

Documentation based Semantic-Aware Log Parsing

With the recent advances of deep learning techniques, there are rapidly growing interests in applying machine learning to log data. As a fundamental part of log analytics, accurate log parsing that transforms raw logs to structured events is critical for subsequent machine learning and data mining tasks. Previous approaches either analyze the source code for parsing or are data-driven such as text clustering. They largely neglect to exploit another widely available and valuable resource, software documentation that provides detailed explanations for the messages, to improve accuracy. In this paper, we propose an approach and system framework to use documentation knowledge for log parsing. With parameter value identification, it not only can improve the parsing accuracy for documented messages but also for undocumented messages. In addition, it can discover the linkages between event templates that are established by sharing parameters and indicate the correlation of the event context.

cs.SE↗

arXiv2021

Probing into gas leakage characteristics of ventilated supercavity through bubbly wake measurement

The stability of ventilated supercavitation is strongly influenced by gas leakage characteristics of the cavity. Here we conduct a systematic investigation of such characteristics under different closure conditions including re-entrant jet (RJ), quad vortex (QV), twin vortex (TV), and pulsating twin vortex (PTV). Using high-speed digital inline holography (DIH), all the individual bubbles shed from the cavity are imaged downstream and are used to quantify the instantaneous gas leakage from the cavity. In general, the supercavity gas leakage exhibits significant fluctuations under all closure types with the instantaneous leakage rate spiking up to 20 times of the ventilation input under RJ and QV closures. However, the magnitude and occurrence rate of such excessive gas leakage vary substantially across different closures, tunnel speeds, and ventilation conditions. Particularly, as the supercavity transitions from RJ, to QV, TV, and PTV with increasing ventilation or decreasing tunnel speed, the relative excessive gas leakage decreases sharply from RJ to QV, plateaus from QV to TV, and drops again from TV to PTV. Correspondingly, the occurrence frequency of such excessive leakage first exhibits a double peak distribution under RJ, migrates to a single peak mode under QV and TV, and eventually transitions to a distribution with a broadened peak at a higher frequency under PTV. These trends can be explained by the flow instabilities associated with three distinct gas leakage mechanisms. Subsequently, two metrics are introduced to quantify the relative change of ventilation needed to compensate for the change of extra gas loss and the predictability of the occurrence of excessive gas leakage, respectively. Based on these metrics, we suggest that the supercavity operating under TV closure with moderate ventilation is optimal for ventilation-based controls of supercavity stability.

physics.flu-dyn↗

arXiv2021

ABCP: Automatic Block-wise and Channel-wise Network Pruning via Joint Search

Currently, an increasing number of model pruning methods are proposed to resolve the contradictions between the computer powers required by the deep learning models and the resource-constrained devices. However, most of the traditional rule-based network pruning methods can not reach a sufficient compression ratio with low accuracy loss and are time-consuming as well as laborious. In this paper, we propose Automatic Block-wise and Channel-wise Network Pruning (ABCP) to jointly search the block-wise and channel-wise pruning action with deep reinforcement learning. A joint sample algorithm is proposed to simultaneously generate the pruning choice of each residual block and the channel pruning ratio of each convolutional layer from the discrete and continuous search space respectively. The best pruning action taking both the accuracy and the complexity of the model into account is obtained finally. Compared with the traditional rule-based pruning method, this pipeline saves human labor and achieves a higher compression ratio with lower accuracy loss. Tested on the mobile robot detection dataset, the pruned YOLOv3 model saves 99.5% FLOPs, reduces 99.5% parameters, and achieves 37.3 times speed up with only 2.8% mAP loss. The results of the transfer task on the sim2real detection dataset also show that our pruned model has much better robustness performance.

cs.CV↗

arXiv2021

Evidence of preferential sweeping during snow settling in atmospheric turbulence

We present a field study of snow settling dynamics based on simultaneous measurements of the atmospheric flow field and snow particle trajectories. Specifically, a super-large-scale particle image velocimetry (SLPIV) system using natural snow particles as tracers is deployed to quantify the velocity field and identify vortex structures in a 22 m $\times$ 39 m field of view centered 18 m above the ground. Simultaneously, we track individual snow particles in a 3 m $\times$ 5 m sample area within the SLPIV using particle tracking velocimetry (PTV). The results reveal the direct linkage among vortex structures in atmospheric turbulence, the spatial distribution of snow particle concentration, and their settling dynamics. In particular, with snow turbulence interaction at near-critical Stokes number, the settling velocity enhancement of snow particles is multifold, and larger than what has been observed in previous field studies. SLPIV measurements show higher concentration of snow particles preferentially located on the downward side of the vortices identified in the atmospheric flow field. PTV, performed on high resolution images around the reconstructed vortices, confirms the latter trend and provides statistical evidence of the acceleration of snow particles, as they move toward the downward side of vortices. Overall, the simultaneous multi-scale particle imaging presented here enables us to directly quantify the salient features of preferential sweeping, supporting it as an underlying mechanism of snow settling enhancement in the atmospheric surface layer.

physics.flu-dyn↗