arXiv subjects

Zhen Li

Publications and source records attributed to Zhen Li.

At least 397 records · Page 22Linked to original sources

arXiv2021

Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision

Pre-trained representations are becoming crucial for many NLP and perception tasks. While representation learning in NLP has transitioned to training on raw text without human annotations, visual and vision-language representations still rely heavily on curated training datasets that are expensive or require expert knowledge. For vision applications, representations are mostly learned using datasets with explicit class labels such as ImageNet or OpenImages. For vision-language, popular datasets like Conceptual Captions, MSCOCO, or CLIP all involve a non-trivial data collection (and cleaning) process. This costly curation process limits the size of datasets and hence hinders the scaling of trained models. In this paper, we leverage a noisy dataset of over one billion image alt-text pairs, obtained without expensive filtering or post-processing steps in the Conceptual Captions dataset. A simple dual-encoder architecture learns to align visual and language representations of the image and text pairs using a contrastive loss. We show that the scale of our corpus can make up for its noise and leads to state-of-the-art representations even with such a simple learning scheme. Our visual representation achieves strong performance when transferred to classification tasks such as ImageNet and VTAB. The aligned visual and language representations enables zero-shot image classification and also set new state-of-the-art results on Flickr30K and MSCOCO image-text retrieval benchmarks, even when compared with more sophisticated cross-attention models. The representations also enable cross-modality search with complex text and text + image queries.

cs.CV↗

arXiv2021

CARLS: Cross-platform Asynchronous Representation Learning System

In this work, we propose CARLS, a novel framework for augmenting the capacity of existing deep learning frameworks by enabling multiple components -- model trainers, knowledge makers and knowledge banks -- to concertedly work together in an asynchronous fashion across hardware platforms. The proposed CARLS is particularly suitable for learning paradigms where model training benefits from additional knowledge inferred or discovered during training, such as node embeddings for graph neural networks or reliable pseudo labels from model predictions. We also describe three learning paradigms -- semi-supervised learning, curriculum learning and multimodal learning -- as examples that can be scaled up efficiently by CARLS. One version of CARLS has been open-sourced and available for download at: https://github.com/tensorflow/neural-structured-learning/tree/master/research/carls

cs.LG↗

arXiv2021

Combining Supervised and Un-supervised Learning for Automatic Citrus Segmentation

Citrus segmentation is a key step of automatic citrus picking. While most current image segmentation approaches achieve good segmentation results by pixel-wise segmentation, these supervised learning-based methods require a large amount of annotated data, and do not consider the continuous temporal changes of citrus position in real-world applications. In this paper, we first train a simple CNN with a small number of labelled citrus images in a supervised manner, which can roughly predict the citrus location from each frame. Then, we extend a state-of-the-art unsupervised learning approach to pre-learn the citrus's potential movements between frames from unlabelled citrus's videos. To take advantages of both networks, we employ the multimodal transformer to combine supervised learned static information and unsupervised learned movement information. The experimental results show that combing both network allows the prediction accuracy reached at 88.3$\%$ IOU and 93.6$\%$ precision, outperforming the original supervised baseline 1.2$\%$ and 2.4$\%$. Compared with most of the existing citrus segmentation methods, our method uses a small amount of supervised data and a large number of unsupervised data, while learning the pixel level location information and the temporal information of citrus changes to enhance the segmentation effect.

cs.CV↗

arXiv2021

VulDeeLocator: A Deep Learning-based Fine-grained Vulnerability Detector

Automatically detecting software vulnerabilities is an important problem that has attracted much attention from the academic research community. However, existing vulnerability detectors still cannot achieve the vulnerability detection capability and the locating precision that would warrant their adoption for real-world use. In this paper, we present a vulnerability detector that can simultaneously achieve a high detection capability and a high locating precision, dubbed Vulnerability Deep learning-based Locator (VulDeeLocator). In the course of designing VulDeeLocator, we encounter difficulties including how to accommodate semantic relations between the definitions of types as well as macros and their uses across files, how to accommodate accurate control flows and variable define-use relations, and how to achieve high locating precision. We solve these difficulties by using two innovative ideas: (i) leveraging intermediate code to accommodate extra semantic information, and (ii) using the notion of granularity refinement to pin down locations of vulnerabilities. When applied to 200 files randomly selected from three real-world software products, VulDeeLocator detects 18 confirmed vulnerabilities (i.e., true-positives). Among them, 16 vulnerabilities correspond to known vulnerabilities; the other two are not reported in the National Vulnerability Database (NVD) but have been "silently" patched by the vendor of Libav when releasing newer versions.

cs.CR↗

arXiv2021

Temporal Modulation Network for Controllable Space-Time Video Super-Resolution

Space-time video super-resolution (STVSR) aims to increase the spatial and temporal resolutions of low-resolution and low-frame-rate videos. Recently, deformable convolution based methods have achieved promising STVSR performance, but they could only infer the intermediate frame pre-defined in the training stage. Besides, these methods undervalued the short-term motion cues among adjacent frames. In this paper, we propose a Temporal Modulation Network (TMNet) to interpolate arbitrary intermediate frame(s) with accurate high-resolution reconstruction. Specifically, we propose a Temporal Modulation Block (TMB) to modulate deformable convolution kernels for controllable feature interpolation. To well exploit the temporal information, we propose a Locally-temporal Feature Comparison (LFC) module, along with the Bi-directional Deformable ConvLSTM, to extract short-term and long-term motion cues in videos. Experiments on three benchmark datasets demonstrate that our TMNet outperforms previous STVSR methods. The code is available at https://github.com/CS-GangXu/TMNet.

cs.CV↗

arXiv2021

PointLIE: Locally Invertible Embedding for Point Cloud Sampling and Recovery

Point Cloud Sampling and Recovery (PCSR) is critical for massive real-time point cloud collection and processing since raw data usually requires large storage and computation. In this paper, we address a fundamental problem in PCSR: How to downsample the dense point cloud with arbitrary scales while preserving the local topology of discarding points in a case-agnostic manner (i.e. without additional storage for point relationship)? We propose a novel Locally Invertible Embedding for point cloud adaptive sampling and recovery (PointLIE). Instead of learning to predict the underlying geometry details in a seemingly plausible manner, PointLIE unifies point cloud sampling and upsampling to one single framework through bi-directional learning. Specifically, PointLIE recursively samples and adjusts neighboring points on each scale. Then it encodes the neighboring offsets of sampled points to a latent space and thus decouples the sampled points and the corresponding local geometric relationship. Once the latent space is determined and that the deep model is optimized, the recovery process could be conducted by passing the recover-pleasing sampled points and a randomly-drawn embedding to the same network through an invertible operation. Such a scheme could guarantee the fidelity of dense point recovery from sampled points. Extensive experiments demonstrate that the proposed PointLIE outperforms state-of-the-arts both quantitatively and qualitatively. Our code is released through https://github.com/zwb0/PointLIE.

cs.CV↗

arXiv2021

Interactive Knowledge Distillation

Knowledge distillation is a standard teacher-student learning framework to train a light-weight student network under the guidance of a well-trained large teacher network. As an effective teaching strategy, interactive teaching has been widely employed at school to motivate students, in which teachers not only provide knowledge but also give constructive feedback to students upon their responses, to improve their learning performance. In this work, we propose an InterActive Knowledge Distillation (IAKD) scheme to leverage the interactive teaching strategy for efficient knowledge distillation. In the distillation process, the interaction between teacher and student networks is implemented by a swapping-in operation: randomly replacing the blocks in the student network with the corresponding blocks in the teacher network. In the way, we directly involve the teacher's powerful feature transformation ability to largely boost the student's performance. Experiments with typical settings of teacher-student networks demonstrate that the student networks trained by our IAKD achieve better performance than those trained by conventional knowledge distillation methods on diverse image classification datasets.

cs.CV↗

arXiv2021

A Holistic Representation Guided Attention Network for Scene Text Recognition

Reading irregular scene text of arbitrary shape in natural images is still a challenging problem, despite the progress made recently. Many existing approaches incorporate sophisticated network structures to handle various shapes, use extra annotations for stronger supervision, or employ hard-to-train recurrent neural networks for sequence modeling. In this work, we propose a simple yet strong approach for scene text recognition. With no need to convert input images to sequence representations, we directly connect two-dimensional CNN features to an attention-based sequence decoder which guided by holistic representation. The holistic representation can guide the attention-based decoder focus on more accurate area. As no recurrent module is adopted, our model can be trained in parallel. It achieves 1.5x to 9.4x acceleration to backward pass and 1.3x to 7.9x acceleration to forward pass, compared with the RNN counterparts. The proposed model is trained with only word-level annotations. With this simple design, our method achieves state-of-the-art or competitive recognition performance on the evaluated regular and irregular scene text benchmark datasets.

cs.CV↗

arXiv2021

Free-form Description Guided 3D Visual Graph Network for Object Grounding in Point Cloud

3D object grounding aims to locate the most relevant target object in a raw point cloud scene based on a free-form language description. Understanding complex and diverse descriptions, and lifting them directly to a point cloud is a new and challenging topic due to the irregular and sparse nature of point clouds. There are three main challenges in 3D object grounding: to find the main focus in the complex and diverse description; to understand the point cloud scene; and to locate the target object. In this paper, we address all three challenges. Firstly, we propose a language scene graph module to capture the rich structure and long-distance phrase correlations. Secondly, we introduce a multi-level 3D proposal relation graph module to extract the object-object and object-scene co-occurrence relationships, and strengthen the visual features of the initial proposals. Lastly, we develop a description guided 3D visual graph module to encode global contexts of phrases and proposals by a nodes matching strategy. Extensive experiments on challenging benchmark datasets (ScanRefer and Nr3D) show that our algorithm outperforms existing state-of-the-art. Our code is available at https://github.com/PNXD/FFL-3DOG.

cs.CV↗

arXiv2021

Quantum emission assisted by energy landscape modification in pentacene-decorated carbon nanotubes

Photoluminescent carbon nanotubes are expected to become versatile room-temperature single-photon sources that have applications in quantum information processing. Quantum emission from carbon nanotubes is often induced by localization of excitons or exciton-exciton annihilation. Here, we modify the local energy landscape of excitons by decorating nanoscale pentacene particles onto air-suspended single-walled carbon nanotubes. Directional exciton diffusion from the undecorated region to the decorated site is demonstrated, suggesting exciton trapping induced by local dielectric screening from pentacene particles. Photoluminescence and photon correlation measurements on a representative carbon nanotube reveal enhanced exciton-exciton annihilation and single-photon emission at room temperature. Pentacene particles are shown to promote strong photon antibunching at the decorated site, indicating that noncovalent functionalization using molecules can be an effective approach for energy landscape modification and quantum emission in carbon nanotubes.

cond-mat.mes-hall↗

arXiv2021

Multiscale Parareal Algorithm for Long-Time Mesoscopic Simulations of Microvascular Blood Flow in Zebrafish

Various biological processes such as transport of oxygen and nutrients, thrombus formation, vascular angiogenesis and remodeling are related to cellular/subcellular level biological processes, where mesoscopic simulations resolving detailed cell dynamics provide a key to understanding and identifying the cellular basis of disease. To break this bottleneck and achieve a biologically meaningful timescale, we propose a multiscale parareal algorithm in which a continuum-based solver supervises a mesoscopic simulation in the time-domain. Using an iterative prediction-correction strategy, the parallel-in-time mesoscopic simulation supervised by its continuum-based counterpart can converge fast. The effectiveness of the proposed method is first verified in a time-dependent flow with a sinusoidal flowrate through a Y-shaped bifurcation channel. Physical quantities of interest including velocity, wall shear stress and flowrate are computed to compare against those of reference solutions, showing a less than 1% relative error on flowrate in the Newtonian flow and a less than 3\% relative error in the non-Newtonian blood flow. The proposed method is then applied to a large-scale mesoscopic simulation of microvessel blood flow in a zebrafish hindbrain for temporal acceleration. The time-dependent blood flow from heartbeats in this realistic vascular network of zebrafish hindbrain is simulated using dissipative particle dynamics as the mesoscopic model, which is supervised by a one-dimensional blood flow model (continuum-based model) in multiple temporal sub-domains. The computational analysis shows that the resulting microvessel blood flow converges to the reference solution after only two iterations. The proposed method is suitable for long-time mesoscopic simulations with complex fluids and geometries.

physics.flu-dyn↗

arXiv2021

SySeVR: A Framework for Using Deep Learning to Detect Software Vulnerabilities

The detection of software vulnerabilities (or vulnerabilities for short) is an important problem that has yet to be tackled, as manifested by the many vulnerabilities reported on a daily basis. This calls for machine learning methods for vulnerability detection. Deep learning is attractive for this purpose because it alleviates the requirement to manually define features. Despite the tremendous success of deep learning in other application domains, its applicability to vulnerability detection is not systematically understood. In order to fill this void, we propose the first systematic framework for using deep learning to detect vulnerabilities in C/C++ programs with source code. The framework, dubbed Syntax-based, Semantics-based, and Vector Representations (SySeVR), focuses on obtaining program representations that can accommodate syntax and semantic information pertinent to vulnerabilities. Our experiments with 4 software products demonstrate the usefulness of the framework: we detect 15 vulnerabilities that are not reported in the National Vulnerability Database. Among these 15 vulnerabilities, 7 are unknown and have been reported to the vendors, and the other 8 have been "silently" patched by the vendors when releasing newer versions of the pertinent software products.

cs.LG↗

arXiv2020

GAKP: GRU Association and Kalman Prediction for Multiple Object Tracking

Multiple Object Tracking (MOT) has been a useful yet challenging task in many real-world applications such as video surveillance, intelligent retail, and smart city. The challenge is how to model long-term temporal dependencies in an efficient manner. Some recent works employ Recurrent Neural Networks (RNN) to obtain good performance, which, however, requires a large amount of training data. In this paper, we proposed a novel tracking method that integrates the auto-tuning Kalman method for prediction and the Gated Recurrent Unit (GRU) and achieves a near-optimum with a small amount of training data. Experimental results show that our new algorithm can achieve competitive performance on the challenging MOT benchmark, and faster and more robust than the state-of-the-art RNN-based online MOT algorithms.

cs.CV↗

arXiv2020

Formal modeling and performance evaluation for hybrid systems:a probabilistic hybrid process algebra-based approach

Probabilistic behavior is omnipresent in computer controlled systems, in particular, so-called safety-critical hybrid systems, because of various reasons, like uncertain environments, or fundamental properties of nature. In this paper, we extend existing hybrid process algebra $ACP_{hs}^{srt}$ with probability without replacing nondeterministic choice operator. In view of some shortcomings in existing approximate probabilistic bisimulation, we relax the constrains and propose a novel approximate probabilistic bisimulation relation. After that, we present a performance evaluation language, CTRML, to reason over probabilistic systems, which extend the results to real number. Along with the specification language, we present a set of algorithms for the evaluation of the language. Additionally, we transfer the hybrid process algebra to probabilistic transition system and show experimental results.

cs.FL↗

arXiv2020

Operator learning for predicting multiscale bubble growth dynamics

Simulating and predicting multiscale problems that couple multiple physics and dynamics across many orders of spatiotemporal scales is a great challenge that has not been investigated systematically by deep neural networks (DNNs). Herein, we develop a framework based on operator regression, the so-called deep operator network (DeepONet), with the long term objective to simplify multiscale modeling by avoiding the fragile and time-consuming "hand-shaking" interface algorithms for stitching together heterogeneous descriptions of multiscale phenomena. To this end, as a first step, we investigate if a DeepONet can learn the dynamics of different scale regimes, one at the deterministic macroscale and the other at the stochastic microscale regime with inherent thermal fluctuations. Specifically, we test the effectiveness and accuracy of DeepONet in predicting multirate bubble growth dynamics, which is described by a Rayleigh-Plesset (R-P) equation at the macroscale and modeled as a stochastic nucleation and cavitation process at the microscale by dissipative particle dynamics (DPD). Taken together, our findings demonstrate that DeepONets can be employed to unify the macroscale and microscale models of the multirate bubble growth problem, hence providing new insight into the role of operator regression via DNNs in tackling realistic multiscale problems and in simplifying modeling with heterogeneous descriptions.

physics.comp-ph↗

arXiv2020

Hierarchically nanostructured thermoelectric materials: Challenges and opportunities for improved power factors

The field of thermoelectric materials has undergone a revolutionary transformation over the last couple of decades as a result of the ability to nanostructure and synthesize myriads of materials and their alloys. The ZT figure of merit, which quantifies the performance of a thermoelectric material has more than doubled after decades of inactivity, reaching values larger than two, consistently across materials and temperatures. Central to this ZT improvement is the drastic reduction in the material thermal conductivity due to the scattering of phonons on the numerous interfaces, boundaries, dislocations, point defects, phases, etc., which are purposely included. In these new generation of nanostructured materials, phonon scattering centers of different sizes and geometrical configurations (atomic, nano- and macro-scale) are formed, which are able to scatter phonons of mean-free-paths across the spectrum. Beyond thermal conductivity reductions, ideas are beginning to emerge on how to use similar hierarchical nanostructuring to achieve power factor improvements. Ways that relax the adverse interdependence of the electrical conductivity and Seebeck coefficient are targeted, which allows power factor improvements. For this, elegant designs are required, that utilize for instance non-uniformities in the underlying nanostructured geometry, non-uniformities in the dopant distribution, or potential barriers that form at boundaries between materials. A few recent reports, both theoretical and experimental, indicate that extremely high power factor values can be achieved, even for the same geometries that also provide ultra-low thermal conductivities. Despite the experimental complications that can arise in having the required control in nanostructure realization, in this colloquium, we aim to demonstrate, mostly theoretically, that it is a very promising path worth exploring.

cond-mat.mtrl-sci↗

arXiv2020

Sparse Single Sweep LiDAR Point Cloud Segmentation via Learning Contextual Shape Priors from Scene Completion

LiDAR point cloud analysis is a core task for 3D computer vision, especially for autonomous driving. However, due to the severe sparsity and noise interference in the single sweep LiDAR point cloud, the accurate semantic segmentation is non-trivial to achieve. In this paper, we propose a novel sparse LiDAR point cloud semantic segmentation framework assisted by learned contextual shape priors. In practice, an initial semantic segmentation (SS) of a single sweep point cloud can be achieved by any appealing network and then flows into the semantic scene completion (SSC) module as the input. By merging multiple frames in the LiDAR sequence as supervision, the optimized SSC module has learned the contextual shape priors from sequential LiDAR data, completing the sparse single sweep point cloud to the dense one. Thus, it inherently improves SS optimization through fully end-to-end training. Besides, a Point-Voxel Interaction (PVI) module is proposed to further enhance the knowledge fusion between SS and SSC tasks, i.e., promoting the interaction of incomplete local geometry of point cloud and complete voxel-wise global structure. Furthermore, the auxiliary SSC and PVI modules can be discarded during inference without extra burden for SS. Extensive experiments confirm that our JS3C-Net achieves superior performance on both SemanticKITTI and SemanticPOSS benchmarks, i.e., 4% and 3% improvement correspondingly.

cs.CV↗

arXiv2020

Progressive Learning Algorithm for Efficient Person Re-Identification

This paper studies the problem of Person Re-Identification (ReID)for large-scale applications. Recent research efforts have been devoted to building complicated part models, which introduce considerably high computational cost and memory consumption, inhibiting its practicability in large-scale applications. This paper aims to develop a novel learning strategy to find efficient feature embeddings while maintaining the balance of accuracy and model complexity. More specifically, we find by enhancing the classical triplet loss together with cross-entropy loss, our method can explore the hard examples and build a discriminant feature embedding yet compact enough for large-scale applications. Our method is carried out progressively using Bayesian optimization, and we call it the Progressive Learning Algorithm (PLA). Extensive experiments on three large-scale datasets show that our PLA is comparable or better than the-state-of-the-arts. Especially, on the challenging Market-1501 dataset, we achieve Rank-1=94.7\%/mAP=89.4\% while saving at least 30\% parameters than strong part models.

cs.CV↗