arXiv subjects

Tao Yang

Publications and source records attributed to Tao Yang.

At least 235 records · Page 13Linked to original sources

arXiv2023

Koos Classification of Vestibular Schwannoma via Image Translation-Based Unsupervised Cross-Modality Domain Adaptation

The Koos grading scale is a classification system for vestibular schwannoma (VS) used to characterize the tumor and its effects on adjacent brain structures. The Koos classification captures many of the characteristics of treatment deci-sions and is often used to determine treatment plans. Although both contrast-enhanced T1 (ceT1) scanning and high-resolution T2 (hrT2) scanning can be used for Koos Classification, hrT2 scanning is gaining interest because of its higher safety and cost-effectiveness. However, in the absence of annotations for hrT2 scans, deep learning methods often inevitably suffer from performance deg-radation due to unsupervised learning. If ceT1 scans and their annotations can be used for unsupervised learning of hrT2 scans, the performance of Koos classifi-cation using unlabeled hrT2 scans will be greatly improved. In this regard, we propose an unsupervised cross-modality domain adaptation method based on im-age translation by transforming annotated ceT1 scans into hrT2 modality and us-ing their annotations to achieve supervised learning of hrT2 modality. Then, the VS and 7 adjacent brain structures related to Koos classification in hrT2 scans were segmented. Finally, handcrafted features are extracted from the segmenta-tion results, and Koos grade is classified using a random forest classifier. The proposed method received rank 1 on the Koos classification task of the Cross-Modality Domain Adaptation (crossMoDA 2022) challenge, with Macro-Averaged Mean Absolute Error (MA-MAE) of 0.2148 for the validation set and 0.26 for the test set.

eess.IV↗

arXiv2023

Synthesizing Realistic Image Restoration Training Pairs: A Diffusion Approach

In supervised image restoration tasks, one key issue is how to obtain the aligned high-quality (HQ) and low-quality (LQ) training image pairs. Unfortunately, such HQ-LQ training pairs are hard to capture in practice, and hard to synthesize due to the complex unknown degradation in the wild. While several sophisticated degradation models have been manually designed to synthesize LQ images from their HQ counterparts, the distribution gap between the synthesized and real-world LQ images remains large. We propose a new approach to synthesizing realistic image restoration training pairs using the emerging denoising diffusion probabilistic model (DDPM). First, we train a DDPM, which could convert a noisy input into the desired LQ image, with a large amount of collected LQ images, which define the target data distribution. Then, for a given HQ image, we synthesize an initial LQ image by using an off-the-shelf degradation model, and iteratively add proper Gaussian noises to it. Finally, we denoise the noisy LQ image using the pre-trained DDPM to obtain the final LQ image, which falls into the target distribution of real-world LQ images. Thanks to the strong capability of DDPM in distribution approximation, the synthesized HQ-LQ image pairs can be used to train robust models for real-world image restoration tasks, such as blind face image restoration and blind image super-resolution. Experiments demonstrated the superiority of our proposed approach to existing degradation models. Code and data will be released.

cs.CV↗

arXiv2023

Development and Evaluation of a Narrow Linewidth Laser System for 171Yb+ E2 Transition

We report the construction and characterization of a narrow-linewidth laser system to interrogate the E2 clock transitions at 436 nm of ytterbium ions trapped in end-cap traps. The 871 nm seed laser at the fundamental frequency is referenced to a 10 cm long notched ULE cavity. The output of the laser system is delivered to a narrow-linewidth femtosecond fiber comb, which has been referenced to an ultrastable 698 nm laser, with a phase noise-canceled fiber link. The beat between the laser and the comb shows a sub-Hz linewidth, and with a stability better than 2E-15@1~100 s. The performance of the self-developed wavelength extension ports at 871 nm of the narrow linewidth erbium-doped fiber comb with single-point frequency-doubling technique is also verified.

physics.atom-ph↗

arXiv2023

Parameter estimation of eccentric gravitational waves with a decihertz observatory and its cosmological implications

Eccentricity of compact binaries can improve the parameter estimation of gravitational waves (GWs). In this paper, we first investigate the parameter estimation of eccentric GWs with decihertz observatory. We consider two scenarios for the configuration of DECIGO, i.e., the one cluster of DECIGO with its design sensitivity and B-DECIGO which also has one cluster but with inferior sensitivity as a comparison. We adopt the Fisher matrix to estimate the parameter errors. By mocking up the typical binaries in GWTC-3, we find a nonvanishing eccentricity can significantly improve the estimation for almost all waveform parameters. In particular, the localization of typical binary black holes (BBH) can achieve $\mathcal{O}(10-10^{3.5})$ factors of improvement when the initial eccentricity $e_0=0.4$ at 0.1 Hz. The precise localization of binary neutron stars (BNS) and neutron star--black hole binaries (NSBH), together with the large improvement of localization of BBH from eccentricity in the mid-band, inspire us to construct the catalogs of golden dark sirens whose host galaxies can be uniquely identified. We find that with only one cluster of DECIGO running 1 year in its design sensitivity, hundreds of golden dark BNS, NSBH, and tens of golden dark BBH can be observed. Eccentricity can greatly increase the population of golden dark BBH from $\sim 7~(e_0=0)$ to $\sim 65~(e_0=0.2)$. Such an increase of population of golden dark BBH events can improve the precision of Hubble constant measurement from 2.06\% to 0.68\%, matter density parameter from 64\% to 16\% in $Λ$CDM model. Through the phenomenological parameterization of GW propagation, the constraints of modified gravity can be improved from 6.2\% to 1.6\%. Our results show the remarkable significance of eccentricity for the detection and parameter estimation of GW events, allowing us to probe the Universe precisely.

gr-qc↗

arXiv2023

Research on road object detection algorithm based on improved YOLOX

Road object detection is an important branch of automatic driving technology, The model with higher detection accuracy is more conducive to the safe driving of vehicles. In road object detection, the omission of small objects and occluded objects is an important problem. therefore, reducing the missed rate of the object is of great significance for safe driving. In the work of this paper, based on the YOLOX object detection algorithm to improve, proposes DecIoU boundary box regression loss function to improve the shape consistency of the predicted and real box, and Push Loss is introduced to further optimize the boundary box regression loss function, in order to detect more occluded objects. In addition, the dynamic anchor box mechanism is also used to improve the accuracy of the confidence label, improve the label inaccuracy of object detection model without anchor box. A large number of experiments on KITTI dataset demonstrate the effectiveness of the proposed method, the improved YOLOX-s achieved 88.9% mAP and 91.0% mAR on the KITTI dataset, compared to the baseline version improvements of 2.77% and 4.24%; the improved YOLOX-m achieved 89.1% mAP and 91.4% mAR, compared to the baseline version improvements of 2.30% and 4.10%.

cs.CV↗

arXiv2023

DroidRL: Reinforcement Learning Driven Feature Selection for Android Malware Detection

Due to the completely open-source nature of Android, the exploitable vulnerability of malware attacks is increasing. Machine learning, leading to a great evolution in Android malware detection in recent years, is typically applied in the classification phase. Since the correlation between features is ignored in some traditional ranking-based feature selection algorithms, applying wrapper-based feature selection models is a topic worth investigating. Though considering the correlation between features, wrapper-based approaches are time-consuming for exploring all possible valid feature subsets when processing a large number of Android features. To reduce the computational expense of wrapper-based feature selection, a framework named DroidRL is proposed. The framework deploys DDQN algorithm to obtain a subset of features which can be used for effective malware classification. To select a valid subset of features over a larger range, the exploration-exploitation policy is applied in the model training phase. The recurrent neural network (RNN) is used as the decision network of DDQN to give the framework the ability to sequentially select features. Word embedding is applied for feature representation to enhance the framework's ability to find the semantic relevance of features. The framework's feature selection exhibits high performance without any human intervention and can be ported to other feature selection tasks with minor changes. The experiment results show a significant effect when using the Random Forest as DroidRL's classifier, which reaches 95.6% accuracy with only 24 features selected.

cs.CR↗

arXiv2023

Normalized solutions to a class of $(2,q)$-Laplacian equations

This paper concerns the existence of normalized solutions to a class of $(2,q)$-Laplacian equations in all the possible cases according to the value of $p$ with respect to the critical exponent $2(1+2/N)$. In the $L^2$-subcritical case, we study a global minimization problem and obtain a ground state solution. While in the $L^2$-critical case, we prove several nonexistence results, extended also in the $L^q$-critical case. At last, we derive a ground state and infinitely many radial solutions in the $L^2$-supercritical case. Compared with the classical Schrödinger equation, the $(2,q)$-Laplacian equation possesses a quasi-linear term, which brings in some new difficulties and requires a more subtle analysis technique. Moreover, the vector field $\vec{a}(ξ)=|ξ|^{q-2}ξ$ corresponding to the $q$-Laplacian is not strictly monotone when $q<2$, so we shall consider separately the case $q<2$ and the case $q>2$.

math.AP↗

arXiv2022

Regret and Cumulative Constraint Violation Analysis for Distributed Online Constrained Convex Optimization

This paper considers the distributed online convex optimization problem with time-varying constraints over a network of agents. This is a sequential decision making problem with two sequences of arbitrarily varying convex loss and constraint functions. At each round, each agent selects a decision from the decision set, and then only a portion of the loss function and a coordinate block of the constraint function at this round are privately revealed to this agent. The goal of the network is to minimize the network-wide loss accumulated over time. Two distributed online algorithms with full-information and bandit feedback are proposed. Both dynamic and static network regret bounds are analyzed for the proposed algorithms, and network cumulative constraint violation is used to measure constraint violation, which excludes the situation that strictly feasible constraints can compensate the effects of violated constraints. In particular, we show that the proposed algorithms achieve $\mathcal{O}(T^{\max\{κ,1-κ\}})$ static network regret and $\mathcal{O}(T^{1-κ/2})$ network cumulative constraint violation, where $T$ is the time horizon and $κ\in(0,1)$ is a user-defined trade-off parameter. Moreover, if the loss functions are strongly convex, then the static network regret bound can be reduced to $\mathcal{O}(T^κ)$. Finally, numerical simulations are provided to illustrate the effectiveness of the theoretical results.

math.OC↗

arXiv2022

Marginal-Certainty-aware Fair Ranking Algorithm

Ranking systems are ubiquitous in modern Internet services, including online marketplaces, social media, and search engines. Traditionally, ranking systems only focus on how to get better relevance estimation. When relevance estimation is available, they usually adopt a user-centric optimization strategy where ranked lists are generated by sorting items according to their estimated relevance. However, such user-centric optimization ignores the fact that item providers also draw utility from ranking systems. It has been shown in existing research that such user-centric optimization will cause much unfairness to item providers, followed by unfair opportunities and unfair economic gains for item providers. To address ranking fairness, many fair ranking methods have been proposed. However, as we show in this paper, these methods could be suboptimal as they directly rely on the relevance estimation without being aware of the uncertainty (i.e., the variance of the estimated relevance). To address this uncertainty, we propose a novel Marginal-Certainty-aware Fair algorithm named MCFair. MCFair jointly optimizes fairness and user utility, while relevance estimation is constantly updated in an online manner. In MCFair, we first develop a ranking objective that includes uncertainty, fairness, and user utility. Then we directly use the gradient of the ranking objective as the ranking score. We theoretically prove that MCFair based on gradients is optimal for the aforementioned ranking objective. Empirically, we find that on semi-synthesized datasets, MCFair is effective and practical and can deliver superior performance compared to state-of-the-art fair ranking methods. To facilitate reproducibility, we release our code https://github.com/Taosheng-ty/WSDM22-MCFair.

cs.IR↗

arXiv2022

Braided crossed category over crossed group-cograded weak Hopf quasigroups

In this paper, we generalizing the main result in Liu[10] to weak Hopf coquasigroups case. We first define and study group-cograded weak Hopf quasigroups, which generalize both group-cograded Hopf quasigroups and weak Hopf group-coalgebras. Then we introduce the notion of p-Yetter-Drinfeld weak quasimodule over group-cograded weak Hopf quasigroups H. If the antipode of H is bijective, we show that the category YDWQ(H) of Yetter-Drinfeld weak quasimodules over H is a crossed category, and the subcategory YD(H) of Yetter-Drinfeld modules is a braided crossed category.

math.RA↗

arXiv2022

Extracting Spatial Interaction Patterns between Urban Road Networks and Mixed Functions

In the field of urban planning, road network system planning is often the first step and the main purpose of urban planning is to create a spatial configuration of different functions such as residence, education, business, etc. Generally speaking, the more mixed the functions of an area has, the more possible its vitality may be. Therefore, in this article, we propose a new framework to study the specific spatial influence patterns of the overall structure and different sub-structures of road networks on the mixed functions. Taking road segment as the basic unit, we characterize mixed functions aggregation of road networks with the number of POIs categories within 100 meters around every road segment. At the same time, on the basis of centrality measurement in graph theory, we use 4 indexes to reflect the characteristics of the urban road network structure, including degree, closeness, betweenness, and length. We conduct our methods and experiments using the road networks and POI data within the 5th ring road of Beijing. The results demonstrate that urban road networks inherently influence the aggregation of urban mixed functions in various patterns and the patterns of road network sub-structures is also quite different. Our study shows that the higher the degree of the road network structure has, the more likely it will attract functions' aggregation. It also reveals that diversified local degree will help gather urban functions. In addition to those, the analysis as well validates the importance of small-grids typology of urban road networks and the closeness to the center of cities.

physics.soc-ph↗

arXiv2022

Eccentricity of Long Inspiraling Compact Binaries Sheds Light on Dark Sirens

The localization and distance inference of gravitational waves are two crucial factors for dark sirens as precise probes of cosmology, astrophysics, and fundamental physics. In this Letter, for the first time we investigate the parameter estimation of gravitational waves emitted by the eccentric compact binaries in the mid-frequency (0.1--10 Hz) band. Based on the configuration of one cluster of DECIGO (B-DECIGO), we simulate five types of typical compact binaries in GWTC-3 with component mass ranging from $\mathcal{O}(1\sim100)~M_{\odot}$. For each type of binaries, we assign discrete eccentricities from 0 to 0.4 at 0.1 Hz in $10^3$ random orientations. The multiple harmonics induced by eccentricity can break the degeneracy between parameters. We find that with eccentricity $e_0=0.4$, these typical binaries can achieve $\mathcal{O}(10^2-10^4)$ improvement for the distance inference in the near face-on orientations, compared to the circular case. More importantly, a nonvanishing eccentricity ($0.01\sim0.4$) can significantly improve the source localization of the typical binary black holes, most by $1.5\sim{3.5}$ orders of magnitude. Our result shows the remarkable significance of eccentricity for dark sirens in the mid-band as precise probes of the Universe.

gr-qc↗

arXiv2022

RePAST: A ReRAM-based PIM Accelerator for Second-order Training of DNN

The second-order training methods can converge much faster than first-order optimizers in DNN training. This is because the second-order training utilizes the inversion of the second-order information (SOI) matrix to find a more accurate descent direction and step size. However, the huge SOI matrices bring significant computational and memory overheads in the traditional architectures like GPU and CPU. On the other side, the ReRAM-based process-in-memory (PIM) technology is suitable for the second-order training because of the following three reasons: First, PIM's computation happens in memory, which reduces data movement overheads; Second, ReRAM crossbars can compute SOI's inversion in $O\left(1\right)$ time; Third, if architected properly, ReRAM crossbars can perform matrix inversion and vector-matrix multiplications which are important to the second-order training algorithms. Nevertheless, current ReRAM-based PIM techniques still face a key challenge for accelerating the second-order training. The existing ReRAM-based matrix inversion circuitry can only support 8-bit accuracy matrix inversion and the computational precision is not sufficient for the second-order training that needs at least 16-bit accurate matrix inversion. In this work, we propose a method to achieve high-precision matrix inversion based on a proven 8-bit matrix inversion (INV) circuitry and vector-matrix multiplication (VMM) circuitry. We design \archname{}, a ReRAM-based PIM accelerator architecture for the second-order training. Moreover, we propose a software mapping scheme for \archname{} to further optimize the performance by fusing VMM and INV crossbar. Experiment shows that \archname{} can achieve an average of 115.8$\times$/11.4$\times$ speedup and 41.9$\times$/12.8$\times$energy saving compared to a GPU counterpart and PipeLayer on large-scale DNNs.

cs.AR↗

arXiv2022

Space-borne atom interferometric gravitational wave detections. Part III. Eccentricity on dark sirens

Eccentricity of the inspiraling compact binaries can greatly improve the distance inference and source localization of dark sirens. In this paper, we continue the research for the space-borne atom interferometric gravitational-wave detector AEDGE and investigate the effects of eccentricity on the dark sirens observed by AEDGE in the mid-band. We simulate five types of typical compact binaries with component mass ranging from $1-100~M_{\odot}$. The largest improvement for both distance inference and localization can be as much as 1.5--3 orders of magnitude. We then construct the catalogs of dark sirens observed by AEDGE in five years. We find eccentricity is crucial to the detection of golden binary black holes (BBH) whose host galaxy can be uniquely identified. With only 5--10 golden dark BBHs one can obtain a 2 percent precision measurement of $H_0$ which is sufficient to arbitrate the Hubble tension. Regardless of eccentricity, AEDGE can also observe tens of golden binary neutron stars (BNS) and neutron star--black hole binaries (NSBH) with unique host galaxies. These golden dark sirens can serve as early warnings for the follow-up observations of gravitational waves in the high frequency band as well as the search of their electromagnetic counterparts. Our results show eccentricity is a crucial factor in the detection, data analysis, and application of GWs with the atom interferometers in the mid-band.

gr-qc↗

arXiv2022

Visual Concepts Tokenization

Obtaining the human-like perception ability of abstracting visual concepts from concrete pixels has always been a fundamental and important target in machine learning research fields such as disentangled representation learning and scene decomposition. Towards this goal, we propose an unsupervised transformer-based Visual Concepts Tokenization framework, dubbed VCT, to perceive an image into a set of disentangled visual concept tokens, with each concept token responding to one type of independent visual concept. Particularly, to obtain these concept tokens, we only use cross-attention to extract visual information from the image tokens layer by layer without self-attention between concept tokens, preventing information leakage across concept tokens. We further propose a Concept Disentangling Loss to facilitate that different concept tokens represent independent visual concepts. The cross-attention and disentangling loss play the role of induction and mutual exclusion for the concept tokens, respectively. Extensive experiments on several popular datasets verify the effectiveness of VCT on the tasks of disentangled representation learning and scene decomposition. VCT achieves the state of the art results by a large margin.

cs.CV↗

arXiv2022

AD-DROP: Attribution-Driven Dropout for Robust Language Model Fine-Tuning

Fine-tuning large pre-trained language models on downstream tasks is apt to suffer from overfitting when limited training data is available. While dropout proves to be an effective antidote by randomly dropping a proportion of units, existing research has not examined its effect on the self-attention mechanism. In this paper, we investigate this problem through self-attention attribution and find that dropping attention positions with low attribution scores can accelerate training and increase the risk of overfitting. Motivated by this observation, we propose Attribution-Driven Dropout (AD-DROP), which randomly discards some high-attribution positions to encourage the model to make predictions by relying more on low-attribution positions to reduce overfitting. We also develop a cross-tuning strategy to alternate fine-tuning and AD-DROP to avoid dropping high-attribution positions excessively. Extensive experiments on various benchmarks show that AD-DROP yields consistent improvements over baselines. Analysis further confirms that AD-DROP serves as a strategic regularizer to prevent overfitting during fine-tuning.

cs.CL↗

arXiv2022

Communication Compression for Distributed Nonconvex Optimization

This paper considers distributed nonconvex optimization with the cost functions being distributed over agents. Noting that information compression is a key tool to reduce the heavy communication load for distributed algorithms as agents iteratively communicate with neighbors, we propose three distributed primal--dual algorithms with compressed communication. The first two algorithms are applicable to a general class of compressors with bounded relative compression error and the third algorithm is suitable for two general classes of compressors with bounded absolute compression error. We show that the proposed distributed algorithms with compressed communication have comparable convergence properties as state-of-the-art algorithms with exact communication. Specifically, we show that they can find first-order stationary points with sublinear convergence rate $\mathcal{O}(1/T)$ when each local cost function is smooth, where $T$ is the total number of iterations, and find global optima with linear convergence rate under an additional condition that the global cost function satisfies the Polyak--Łojasiewicz condition. Numerical simulations are provided to illustrate the effectiveness of the theoretical results.

math.OC↗

arXiv2022

Doubly-Irregular Repeat-Accumulate Codes over Integer Rings for Multi-user Communications

Structured codes based on lattices were shown to provide enlarged capacity for multi-user communication networks. In this paper, we study capacity-approaching irregular repeat accumulate (IRA) codes over integer rings $\mathbb{Z}_{2^{m}}$ for $2^m$-PAM signaling, $m=1,2,\cdots$. Such codes feature the property that the integer sum of $K$ codewords belongs to the extended codebook (or lattice) w.r.t. the base code. With it, \emph{% structured binning} can be utilized and the gains promised in lattice based network information theory can be materialized in practice. In designing IRA ring codes, we first analyze the effect of zero-divisors of integer ring on the iterative belief-propagation (BP) decoding, and show the invalidity of symmetric Gaussian approximation. Then we propose a doubly IRA (D-IRA) ring code structure, consisting of \emph{irregular multiplier distribution} and \emph{irregular node-degree distribution}, that can restore the symmetry and optimize the BP decoding threshold. For point-to-point AWGN channel with $% 2^m $-PAM inputs, D-IRA ring codes perform as low as 0.29 dB to the capacity limits, outperforming existing bit-interleaved coded-modulation (BICM) and IRA modulation codes over GF($2^m$). We then proceed to design D-IRA ring codes for two important multi-user communication setups, namely compute-forward (CF) and dirty paper coding (DPC), with $2^m$-PAM signaling. With it, a physical-layer network coding scheme yields a gap to the CF limit by 0.24 dB, and a simple linear DPC scheme exhibits a gap to the capacity by 0.91 dB.

cs.IT↗