arXiv subjects

Feng Liu

Publications and source records attributed to Feng Liu.

At least 469 records · Page 26Linked to original sources

arXiv2022

Affine Transformation Edited and Refined Deep Neural Network for Quantitative Susceptibility Mapping

Deep neural networks have demonstrated great potential in solving dipole inversion for Quantitative Susceptibility Mapping (QSM). However, the performances of most existing deep learning methods drastically degrade with mismatched sequence parameters such as acquisition orientation and spatial resolution. We propose an end-to-end AFfine Transformation Edited and Refined (AFTER) deep neural network for QSM, which is robust against arbitrary acquisition orientation and spatial resolution up to 0.6 mm isotropic at the finest. The AFTER-QSM neural network starts with a forward affine transformation layer, followed by an Unet for dipole inversion, then an inverse affine transformation layer, followed by a Residual Dense Network (RDN) for QSM refinement. Simulation and in-vivo experiments demonstrated that the proposed AFTER-QSM network architecture had excellent generalizability. It can successfully reconstruct susceptibility maps from highly oblique and anisotropic scans, leading to the best image quality assessments in simulation tests and suppressed streaking artifacts and noise levels for in-vivo experiments compared with other methods. Furthermore, ablation studies showed that the RDN refinement network significantly reduced image blurring and susceptibility underestimation due to affine transformations. In addition, the AFTER-QSM network substantially shortened the reconstruction time from minutes using conventional methods to only a few seconds.

physics.med-ph↗

arXiv2022

Topological field-effect transistor with quantized ON/OFF conductance of helical/chiral dislocation states

Topology is a key ingredient driving the emergence of quantum devices. Topological field-effect transistor (TFET) has been proposed to outperform the conventional FET by replacing the ON state with topology-protected quantized conductance, while the OFF state remains the same normal insulating characteristics and hence bears similar drawbacks. Here, we demonstrate a proof-of-concept TFET having both ON and OFF quantized conductance, by switching between helical and chiral topological screw dislocation (SD) states in three-dimensional topological insulators. A pair of SDs are configured with one acting as channel and the other as gate controlled by local magnetic field. A reversible field-switching is achieved with the ON and OFF conductance of $2e^2/h$ and $e^2/h$, respectively, as shown by tight-binding quantum transport calculations. Furthermore, BaBiO$_3$ is shown as a candidate material having the desired topological SD states, based on first-principles calculations. Our findings open a new route to high-fidelity topological quantum devices.

cond-mat.mtrl-sci↗

arXiv2022

Semiconductor membranes for electrostatic exciton trapping in optically addressable quantum transport devices

Combining the capabilities of gate defined quantum transport devices in GaAs-based heterostructures and of optically addressed self-assembled quantum dots could open broad perspectives for new devices and functionalities. For example, interfacing stationary solid-state qubits with photonic quantum states would open a new pathway towards the realization of a quantum network with extended quantum processing capacity in each node. While gated devices allow very flexible confinement of electrons or holes, the confinement of excitons without some element of self-assembly is much harder. To address this limitation, we introduce a technique to realize exciton traps in quantum wells via local electric fields by thinning a heterostructure down to a 220 nm thick membrane. We show that mobilities over $1 \times 10^{6}$ cm$^{2}$V$^{-1}$s$^{-1}$ can be retained and that quantum point contacts and Coulomb oscillations can be observed on this structure, which implies that the thinning does not compromise the heterostructure quality. Furthermore, the local lowering of the exciton energy via the quantum-confined Stark effect is confirmed, thus forming exciton traps. These results lay the technological foundations for devices like single photon sources, spin photon interfaces and eventually quantum network nodes in GaAs quantum wells, realized entirely with a top-down fabrication process.

cond-mat.mes-hall↗

arXiv2022

Watermarking for Out-of-distribution Detection

Out-of-distribution (OOD) detection aims to identify OOD data based on representations extracted from well-trained deep models. However, existing methods largely ignore the reprogramming property of deep models and thus may not fully unleash their intrinsic strength: without modifying parameters of a well-trained deep model, we can reprogram this model for a new purpose via data-level manipulation (e.g., adding a specific feature perturbation to the data). This property motivates us to reprogram a classification model to excel at OOD detection (a new task), and thus we propose a general methodology named watermarking in this paper. Specifically, we learn a unified pattern that is superimposed onto features of original data, and the model's detection capability is largely boosted after watermarking. Extensive experiments verify the effectiveness of watermarking, demonstrating the significance of the reprogramming property of deep models in OOD detection.

cs.LG↗

arXiv2022

Mix and Reason: Reasoning over Semantic Topology with Data Mixing for Domain Generalization

Domain generalization (DG) enables generalizing a learning machine from multiple seen source domains to an unseen target one. The general objective of DG methods is to learn semantic representations that are independent of domain labels, which is theoretically sound but empirically challenged due to the complex mixture of common and domain-specific factors. Although disentangling the representations into two disjoint parts has been gaining momentum in DG, the strong presumption over the data limits its efficacy in many real-world scenarios. In this paper, we propose Mix and Reason (\mire), a new DG framework that learns semantic representations via enforcing the structural invariance of semantic topology. \mire\ consists of two key components, namely, Category-aware Data Mixing (CDM) and Adaptive Semantic Topology Refinement (ASTR). CDM mixes two images from different domains in virtue of activation maps generated by two complementary classification losses, making the classifier focus on the representations of semantic objects. ASTR introduces relation graphs to represent semantic topology, which is progressively refined via the interactions between local feature aggregation and global cross-domain relational reasoning. Experiments on multiple DG benchmarks validate the effectiveness and robustness of the proposed \mire.

cs.CV↗

arXiv2022

Excitonic Condensate in Flat Valence and Conduction Bands of Opposite Chirality

Excitonic Bose-Einstein condensation (EBEC) has drawn increasing attention recently with the emergence of 2D materials. A general criterion for EBEC, as expected in an excitonic insulator (EI) state, is to have negative exciton formation energies in a semiconductor. Here, using exact diagonalization of multi-exciton Hamiltonian modelled in a diatomic Kagome lattice, we demonstrate that the negative exciton formation energies are only a prerequisite but insufficient condition for realizing an EI. By a comparative study between the cases of both a conduction and valence flat bands (FBs) versus that of a parabolic conduction band, we further show that the presence and increased FB contribution to exciton formation provide an attractive avenue to stabilize the EBEC, as confirmed by calculations and analyses of multi-exciton energies, wave functions and reduced density matrices. Our results warrant a similar many-exciton analysis for other known/new candidates of EIs, and demonstrate the FBs of opposite parity as a unique platform for studying exciton physics, paving the way to material realization of spinor BEC and spin-superfluidity.

cond-mat.str-el↗

arXiv2022

A Perceptual Quality Metric for Video Frame Interpolation

Research on video frame interpolation has made significant progress in recent years. However, existing methods mostly use off-the-shelf metrics to measure the quality of interpolation results with the exception of a few methods that employ user studies, which is time-consuming. As video frame interpolation results often exhibit unique artifacts, existing quality metrics sometimes are not consistent with human perception when measuring the interpolation results. Some recent deep learning-based perceptual quality metrics are shown more consistent with human judgments, but their performance on videos is compromised since they do not consider temporal information. In this paper, we present a dedicated perceptual quality metric for measuring video frame interpolation results. Our method learns perceptual features directly from videos instead of individual frames. It compares pyramid features extracted from video frames and employs Swin Transformer blocks-based spatio-temporal modules to extract spatio-temporal information. To train our metric, we collected a new video frame interpolation quality assessment dataset. Our experiments show that our dedicated quality metric outperforms state-of-the-art methods when measuring video frame interpolation results. Our code and model are made publicly available at \url{https://github.com/hqqxyy/VFIPS}.

cs.CV↗

arXiv2022

A Uniform Representation Learning Method for OCT-based Fingerprint Presentation Attack Detection and Reconstruction

The technology of optical coherence tomography (OCT) to fingerprint imaging opens up a new research potential for fingerprint recognition owing to its ability to capture depth information of the skin layers. Developing robust and high security Automated Fingerprint Recognition Systems (AFRSs) are possible if the depth information can be fully utilized. However, in existing studies, Presentation Attack Detection (PAD) and subsurface fingerprint reconstruction based on depth information are treated as two independent branches, resulting in high computation and complexity of AFRS building.Thus, this paper proposes a uniform representation model for OCT-based fingerprint PAD and subsurface fingerprint reconstruction. Firstly, we design a novel semantic segmentation network which only trained by real finger slices of OCT-based fingerprints to extract multiple subsurface structures from those slices (also known as B-scans). The latent codes derived from the network are directly used to effectively detect the PA since they contain abundant subsurface biological information, which is independent with PA materials and has strong robustness for unknown PAs. Meanwhile, the segmented subsurface structures are adopted to reconstruct multiple subsurface 2D fingerprints. Recognition can be easily achieved by using existing mature technologies based on traditional 2D fingerprints. Extensive experiments are carried on our own established database, which is the largest public OCT-based fingerprint database with 2449 volumes. In PAD task, our method can improve 0.33% Acc from the state-of-the-art method. For reconstruction performance, our method achieves the best performance with 0.834 mIOU and 0.937 PA. By comparing with the recognition performance on surface 2D fingerprints, the effectiveness of our proposed method on high quality subsurface fingerprint reconstruction is further proved.

cs.CV↗

arXiv2022

Two-Stage Submodular Optimization of Dynamic Thermal Rating for Risk Mitigation Considering Placement and Operation Schedule

Cascading failure causes a major risk to society currently. To effectively mitigate the risk, dynamic thermal rating (DTR) technique can be applied as a cost-effective strategy to exploit potential transmission capability. From the perspectives of service life and Braess paradox, it is important and challenging to jointly optimize the DTR placement and operation schedule for changing system state, which is a two-stage combinatorial problem with only discrete variables, suffering from no approximation guarantee and dimension curse only based on traditional models. Thus, the present work proposes a novel two-stage submodular optimization (TSSO) of DTR for risk mitigation considering placement and operation schedule. Specifically, it optimizes DTR placement with proper redundancy in first stage, and then determines the corresponding DTR operation for each system state in second stage. Under the condition of the Markov and submodular features in sub-function of risk mitigation, the submodularity of total objective function of TSSO can be proven for the first time. Based on this, a state-of-the-art efficient solving algorithm is developed that can provide a better approximation guarantee than previous studies by coordinating the separate curvature and error form. The performance of the proposed algorithms is verified by case results.

eess.SY↗

arXiv2022

Meta Discovery: Learning to Discover Novel Classes given Very Limited Data

In novel class discovery (NCD), we are given labeled data from seen classes and unlabeled data from unseen classes, and we train clustering models for the unseen classes. However, the implicit assumptions behind NCD are still unclear. In this paper, we demystify assumptions behind NCD and find that high-level semantic features should be shared among the seen and unseen classes. Based on this finding, NCD is theoretically solvable under certain assumptions and can be naturally linked to meta-learning that has exactly the same assumption as NCD. Thus, we can empirically solve the NCD problem by meta-learning algorithms after slight modifications. This meta-learning-based methodology significantly reduces the amount of unlabeled data needed for training and makes it more practical, as demonstrated in experiments. The use of very limited data is also justified by the application scenario of NCD: since it is unnatural to label only seen-class data, NCD is sampling instead of labeling in causality. Therefore, unseen-class data should be collected on the way of collecting seen-class data, which is why they are novel and first need to be clustered.

cs.LG↗

arXiv2022

TOHAN: A One-step Approach towards Few-shot Hypothesis Adaptation

In few-shot domain adaptation (FDA), classifiers for the target domain are trained with accessible labeled data in the source domain (SD) and few labeled data in the target domain (TD). However, data usually contain private information in the current era, e.g., data distributed on personal phones. Thus, the private information will be leaked if we directly access data in SD to train a target-domain classifier (required by FDA methods). In this paper, to thoroughly prevent the privacy leakage in SD, we consider a very challenging problem setting, where the classifier for the TD has to be trained using few labeled target data and a well-trained SD classifier, named few-shot hypothesis adaptation (FHA). In FHA, we cannot access data in SD, as a result, the private information in SD will be protected well. To this end, we propose a target orientated hypothesis adaptation network (TOHAN) to solve the FHA problem, where we generate highly-compatible unlabeled data (i.e., an intermediate domain) to help train a target-domain classifier. TOHAN maintains two deep networks simultaneously, where one focuses on learning an intermediate domain and the other takes care of the intermediate-to-target distributional adaptation and the target-risk minimization. Experimental results show that TOHAN outperforms competitive baselines significantly.

cs.LG↗

arXiv2022

Negative Interatomic Spring Constant Manifested by Topological Phonon Flat Band

Phonons as bosons are different from electrons as fermions. Unlike interatomic electron hopping that can be either positive or negative and further tuned by spin-orbit coupling, interatomic spring constant is positive, or the structure of atomic lattices would be dynamically unstable. Surprisingly, we found that topological phonon flat bands (FBs) can manifest either a positive or negative interatomic spring constant that couples the FB-modes of opposite chirality, as exemplified by first-principles calculations of a 2D material of Kagome-BN. To reveal its physical origin, we first establish a fundamental correspondence between a collective lattice-coupling (CLC) variable of two quasi-particle states (e.g., electronic states or phonon modes) of opposite parity in a periodic lattice with band topology. Topological semimetals arise with zero CLC at special k-points protected by symmetry; while positive and negative CLC at these k-points gives rise to normal and topological insulators, respectively. Then, we show topological FB has a special form of CLC that vanishes at all k-points as characterized by its real-space wave function, and multi-atom FB phonon mode can manifest effectively a negative interatomic spring constant. Our findings shed new light on our fundamental understanding of topology and provide a practical design principle for creating artificial bosonic topological states.

cond-mat.mtrl-sci↗

arXiv2022

Neighborhood Collective Estimation for Noisy Label Identification and Correction

Learning with noisy labels (LNL) aims at designing strategies to improve model performance and generalization by mitigating the effects of model overfitting to noisy labels. The key success of LNL lies in identifying as many clean samples as possible from massive noisy data, while rectifying the wrongly assigned noisy labels. Recent advances employ the predicted label distributions of individual samples to perform noise verification and noisy label correction, easily giving rise to confirmation bias. To mitigate this issue, we propose Neighborhood Collective Estimation, in which the predictive reliability of a candidate sample is re-estimated by contrasting it against its feature-space nearest neighbors. Specifically, our method is divided into two steps: 1) Neighborhood Collective Noise Verification to separate all training samples into a clean or noisy subset, 2) Neighborhood Collective Label Correction to relabel noisy samples, and then auxiliary techniques are used to assist further model optimization. Extensive experiments on four commonly used benchmark datasets, i.e., CIFAR-10, CIFAR-100, Clothing-1M and Webvision-1.0, demonstrate that our proposed method considerably outperforms state-of-the-art methods.

cs.CV↗

arXiv2022

FRT-PAD: Effective Presentation Attack Detection Driven by Face Related Task

The robustness and generalization ability of Presentation Attack Detection (PAD) methods is critical to ensure the security of Face Recognition Systems (FRSs). However, in a real scenario, Presentation Attacks (PAs) are various and it is hard to predict the Presentation Attack Instrument (PAI) species that will be used by the attacker. Existing PAD methods are highly dependent on the limited training set and cannot generalize well to unknown PAI species. Unlike this specific PAD task, other face related tasks trained by huge amount of real faces (e.g. face recognition and attribute editing) can be effectively adopted into different application scenarios. Inspired by this, we propose to trade position of PAD and face related work in a face system and apply the free acquired prior knowledge from face related tasks to solve face PAD, so as to improve the generalization ability in detecting PAs. The proposed method, first introduces task specific features from other face related task, then, we design a Cross-Modal Adapter using a Graph Attention Network (GAT) to re-map such features to adapt to PAD task. Finally, face PAD is achieved by using the hierarchical features from a CNN-based PA detector and the re-mapped features. The experimental results show that the proposed method can achieve significant improvements in the complicated and hybrid datasets, when compared with the state-of-the-art methods. In particular, when training on the datasets OULU-NPU, CASIA-FASD, and Idiap Replay-Attack, we obtain HTER (Half Total Error Rate) of 5.48% for the testing dataset MSU-MFSD, outperforming the baseline by 7.39%.

cs.CV↗

arXiv2022

Centrality and Consistency: Two-Stage Clean Samples Identification for Learning with Instance-Dependent Noisy Labels

Deep models trained with noisy labels are prone to over-fitting and struggle in generalization. Most existing solutions are based on an ideal assumption that the label noise is class-conditional, i.e., instances of the same class share the same noise model, and are independent of features. While in practice, the real-world noise patterns are usually more fine-grained as instance-dependent ones, which poses a big challenge, especially in the presence of inter-class imbalance. In this paper, we propose a two-stage clean samples identification method to address the aforementioned challenge. First, we employ a class-level feature clustering procedure for the early identification of clean samples that are near the class-wise prediction centers. Notably, we address the class imbalance problem by aggregating rare classes according to their prediction entropy. Second, for the remaining clean samples that are close to the ground truth class boundary (usually mixed with the samples with instance-dependent noises), we propose a novel consistency-based classification method that identifies them using the consistency of two classifier heads: the higher the consistency, the larger the probability that a sample is clean. Extensive experiments on several challenging benchmarks demonstrate the superior performance of our method against the state-of-the-art.

cs.CV↗

arXiv2022

Shift-tolerant Perceptual Similarity Metric

Existing perceptual similarity metrics assume an image and its reference are well aligned. As a result, these metrics are often sensitive to a small alignment error that is imperceptible to the human eyes. This paper studies the effect of small misalignment, specifically a small shift between the input and reference image, on existing metrics, and accordingly develops a shift-tolerant similarity metric. This paper builds upon LPIPS, a widely used learned perceptual similarity metric, and explores architectural design considerations to make it robust against imperceptible misalignment. Specifically, we study a wide spectrum of neural network elements, such as anti-aliasing filtering, pooling, striding, padding, and skip connection, and discuss their roles in making a robust metric. Based on our studies, we develop a new deep neural network-based perceptual similarity metric. Our experiments show that our metric is tolerant to imperceptible shifts while being consistent with the human similarity judgment.

cs.CV↗

arXiv2022

Controllable and Guided Face Synthesis for Unconstrained Face Recognition

Although significant advances have been made in face recognition (FR), FR in unconstrained environments remains challenging due to the domain gap between the semi-constrained training datasets and unconstrained testing scenarios. To address this problem, we propose a controllable face synthesis model (CFSM) that can mimic the distribution of target datasets in a style latent space. CFSM learns a linear subspace with orthogonal bases in the style latent space with precise control over the diversity and degree of synthesis. Furthermore, the pre-trained synthesis model can be guided by the FR model, making the resulting images more beneficial for FR model training. Besides, target dataset distributions are characterized by the learned orthogonal bases, which can be utilized to measure the distributional similarity among face datasets. Our approach yields significant performance gains on unconstrained benchmarks, such as IJB-B, IJB-C, TinyFace and IJB-S (+5.76% Rank1).

cs.CV↗

arXiv2022

2D GANs Meet Unsupervised Single-view 3D Reconstruction

Recent research has shown that controllable image generation based on pre-trained GANs can benefit a wide range of computer vision tasks. However, less attention has been devoted to 3D vision tasks. In light of this, we propose a novel image-conditioned neural implicit field, which can leverage 2D supervisions from GAN-generated multi-view images and perform the single-view reconstruction of generic objects. Firstly, a novel offline StyleGAN-based generator is presented to generate plausible pseudo images with full control over the viewpoint. Then, we propose to utilize a neural implicit function, along with a differentiable renderer to learn 3D geometry from pseudo images with object masks and rough pose initializations. To further detect the unreliable supervisions, we introduce a novel uncertainty module to predict uncertainty maps, which remedy the negative effect of uncertain regions in pseudo images, leading to a better reconstruction performance. The effectiveness of our approach is demonstrated through superior single-view 3D reconstruction results of generic objects.

cs.CV↗