arXiv subjects

Siqi Zhang

Publications and source records attributed to Siqi Zhang.

At least 55 records · Page 3Linked to original sources

arXiv2024

MiniVLN: Efficient Vision-and-Language Navigation by Progressive Knowledge Distillation

In recent years, Embodied Artificial Intelligence (Embodied AI) has advanced rapidly, yet the increasing size of models conflicts with the limited computational capabilities of Embodied AI platforms. To address this challenge, we aim to achieve both high model performance and practical deployability. Specifically, we focus on Vision-and-Language Navigation (VLN), a core task in Embodied AI. This paper introduces a two-stage knowledge distillation framework, producing a student model, MiniVLN, and showcasing the significant potential of distillation techniques in developing lightweight models. The proposed method aims to capture fine-grained knowledge during the pretraining phase and navigation-specific knowledge during the fine-tuning phase. Our findings indicate that the two-stage distillation approach is more effective in narrowing the performance gap between the teacher model and the student model compared to single-stage distillation. On the public R2R and REVERIE benchmarks, MiniVLN achieves performance on par with the teacher model while having only about 12% of the teacher model's parameter count.

cs.CV↗

arXiv2024

A Survey of Computation Offloading with Task Types

Computation task offloading plays a crucial role in facilitating computation-intensive applications and edge intelligence, particularly in response to the explosive growth of massive data generation. Various enabling techniques, wireless technologies and mechanisms have already been proposed for task offloading, primarily aimed at improving the quality of services (QoS) for users. While there exists an extensive body of literature on this topic, exploring computation offloading from the standpoint of task types has been relatively underrepresented. This motivates our survey, which seeks to classify the state-of-the-art (SoTA) from the task type point-of-view. To achieve this, a thorough literature review is conducted to reveal the SoTA from various aspects, including architecture, objective, offloading strategy, and task types, with the consideration of task generation. It has been observed that task types are associated with data and have an impact on the offloading process, including elements like resource allocation and task assignment. Building upon this insight, computation offloading is categorized into two groups based on task types: static task-based offloading and dynamic task-based offloading. Finally, a prospective view of the challenges and opportunities in the field of future computation offloading is presented.

cs.DC↗

arXiv2024

Communication-Efficient Gradient Descent-Accent Methods for Distributed Variational Inequalities: Unified Analysis and Local Updates

Distributed and federated learning algorithms and techniques associated primarily with minimization problems. However, with the increase of minimax optimization and variational inequality problems in machine learning, the necessity of designing efficient distributed/federated learning approaches for these problems is becoming more apparent. In this paper, we provide a unified convergence analysis of communication-efficient local training methods for distributed variational inequality problems (VIPs). Our approach is based on a general key assumption on the stochastic estimates that allows us to propose and analyze several novel local training algorithms under a single framework for solving a class of structured non-monotone VIPs. We present the first local gradient descent-accent algorithms with provable improved communication complexity for solving distributed variational inequalities on heterogeneous data. The general algorithmic framework recovers state-of-the-art algorithms and their sharp convergence guarantees when the setting is specialized to minimization or minimax optimization problems. Finally, we demonstrate the strong performance of the proposed algorithms compared to state-of-the-art methods when solving federated minimax optimization problems.

math.OC↗

arXiv2024

Biased Stochastic First-Order Methods for Conditional Stochastic Optimization and Applications in Meta Learning

Conditional stochastic optimization covers a variety of applications ranging from invariant learning and causal inference to meta-learning. However, constructing unbiased gradient estimators for such problems is challenging due to the composition structure. As an alternative, we propose a biased stochastic gradient descent (BSGD) algorithm and study the bias-variance tradeoff under different structural assumptions. We establish the sample complexities of BSGD for strongly convex, convex, and weakly convex objectives under smooth and non-smooth conditions. Our lower bound analysis shows that the sample complexities of BSGD cannot be improved for general convex objectives and nonconvex objectives except for smooth nonconvex objectives with Lipschitz continuous gradient estimator. For this special setting, we propose an accelerated algorithm called biased SpiderBoost (BSpiderBoost) that matches the lower bound complexity. We further conduct numerical experiments on invariant logistic regression and model-agnostic meta-learning to illustrate the performance of BSGD and BSpiderBoost.

math.OC↗

arXiv2024

On using the counting method to constrain the anisotropy of kilonova radiation

A large number of binary neutron star (BNS) mergers are expected to be detected by gravitational wave (GW) detectors and the electromagnetic (EM) counterparts (e.g., kilonovae) of a fraction of these mergers may be detected in multi-bands by large area survey telescopes. For a given number of BNS mergers detected by their GW signals, the expected numbers of their EM counterparts that can be detected by a survey with given selection criteria depend on the kilonova properties, including the anisotropy. In this paper, we investigate whether the anisotropy of kilonova radiation and the kilonova model can be constrained statistically by the counting method, i.e., using the numbers of BNS mergers detected via GW and multi-band EM signals. Adopting simple models for the BNS mergers, afterglows, and a simple two (blue and red)-component model for kilonovae, we generate mock samples for GW detected BNS mergers, their associated kilonovae and afterglows detected in multi-bands. By assuming some criteria for searching the EM counterparts, we simulate the observations of these EM counterparts and obtain the EM observed samples in different bands. With the numbers of BNS mergers detected by GW detectors and EM survey telescopes in different bands, we show that the anisotropy of kilonova radiation and the kilonova model can be well constrained by using the Bayesian analysis. Our results suggest that the anisotropy of kilonova radiation may be demographically and globally constrained by simply using the detection numbers of BNS mergers by GW detectors and EM survey telescopes in multi-bands.

astro-ph.HE↗

arXiv2023

On the detection of the electromagnetic counterparts from lensed gravitational wave events by binary neutron star mergers

Future ground-based gravitational wave (GW) detectors, i.e., Einstein telescope (ET) and Cosmic Explorer (CE), are expected to detect a significant number of lensed binary neutron star (BNS) mergers, which may provide a unique tool to probe cosmology. In this paper, we investigate the detectability of the optical/infrared electromagnetic (EM) counterparts (kilonovae/afterglows) from these lensed BNS mergers by future GW detectors and EM telescopes using simple kilonova, afterglow, and lens models. ET and CE are expected to detect $\sim5.32^{+26.1}_{-5.10}$ and $67.3^{+332}_{-64.7}$ lensed BNS mergers per year. We find that the EM counterparts associated with all these mergers will be detectable by an all sky-survey in the H-band with the limiting magnitude $m_{\textrm{lim}}\gtrsim27$, while the detectable fraction is $\lesssim0.4\%$ in the g-/z-band if with $m_{\textrm{lim}}\lesssim24$. Generally it is more efficient to search the lensed EM counterparts by adopting the infrared bands than the optical/UV bands with the same $m_{\textrm{lim}}$. Future telescopes like Vera C. Rubin Observatory, China Space Station Telescope, and Euclid can hardly detect the EM counterparts of even one lensed BNS merger. Roman Space Telescope (RST) and James Webb Space Telescope (JWST) have the capability to detect about a few or more such events per year. Moreover, the time delays and separations between the lensed image pairs are typically in the ranges from minutes to months and from $0.1$ to $1$\,arcsec, suggesting that both the GW and EM images of most lensed BNS mergers can be well resolved by not only CE/ET in the time domain but also RST/JWST spatially.

astro-ph.HE↗

arXiv2023

FIT: Frequency-based Image Translation for Domain Adaptive Object Detection

Domain adaptive object detection (DAOD) aims to adapt the detector from a labelled source domain to an unlabelled target domain. In recent years, DAOD has attracted massive attention since it can alleviate performance degradation due to the large shift of data distributions in the wild. To align distributions between domains, adversarial learning is widely used in existing DAOD methods. However, the decision boundary for the adversarial domain discriminator may be inaccurate, causing the model biased towards the source domain. To alleviate this bias, we propose a novel Frequency-based Image Translation (FIT) framework for DAOD. First, by keeping domain-invariant frequency components and swapping domain-specific ones, we conduct image translation to reduce domain shift at the input level. Second, hierarchical adversarial feature learning is utilized to further mitigate the domain gap at the feature level. Finally, we design a joint loss to train the entire network in an end-to-end manner without extra training to obtain translated images. Extensive experiments on three challenging DAOD benchmarks demonstrate the effectiveness of our method.

cs.CV↗

arXiv2023

Refined Pseudo labeling for Source-free Domain Adaptive Object Detection

Domain adaptive object detection (DAOD) assumes that both labeled source data and unlabeled target data are available for training, but this assumption does not always hold in real-world scenarios. Thus, source-free DAOD is proposed to adapt the source-trained detectors to target domains with only unlabeled target data. Existing source-free DAOD methods typically utilize pseudo labeling, where the performance heavily relies on the selection of confidence threshold. However, most prior works adopt a single fixed threshold for all classes to generate pseudo labels, which ignore the imbalanced class distribution, resulting in biased pseudo labels. In this work, we propose a refined pseudo labeling framework for source-free DAOD. First, to generate unbiased pseudo labels, we present a category-aware adaptive threshold estimation module, which adaptively provides the appropriate threshold for each category. Second, to alleviate incorrect box regression, a localization-aware pseudo label assignment strategy is introduced to divide labels into certain and uncertain ones and optimize them separately. Finally, extensive experiments on four adaptation tasks demonstrate the effectiveness of our method.

cs.CV↗

arXiv2023

Unseen Object Instance Segmentation with Fully Test-time RGB-D Embeddings Adaptation

Segmenting unseen objects is a crucial ability for the robot since it may encounter new environments during the operation. Recently, a popular solution is leveraging RGB-D features of large-scale synthetic data and directly applying the model to unseen real-world scenarios. However, the domain shift caused by the sim2real gap is inevitable, posing a crucial challenge to the segmentation model. In this paper, we emphasize the adaptation process across sim2real domains and model it as a learning problem on the BatchNorm parameters of a simulation-trained model. Specifically, we propose a novel non-parametric entropy objective, which formulates the learning objective for the test-time adaptation in an open-world manner. Then, a cross-modality knowledge distillation objective is further designed to encourage the test-time knowledge transfer for feature enhancement. Our approach can be efficiently implemented with only test images, without requiring annotations or revisiting the large-scale synthetic training data. Besides significant time savings, the proposed method consistently improves segmentation results on the overlap and boundary metrics, achieving state-of-the-art performance on unseen object instance segmentation.

cs.CV↗

arXiv2023

Generalization Bounds of Nonconvex-(Strongly)-Concave Stochastic Minimax Optimization

This paper takes an initial step to systematically investigate the generalization bounds of algorithms for solving nonconvex-(strongly)-concave (NC-SC/NC-C) stochastic minimax optimization measured by the stationarity of primal functions. We first establish algorithm-agnostic generalization bounds via uniform convergence between the empirical minimax problem and the population minimax problem. The sample complexities for achieving $ε$-generalization are $\tilde{\mathcal{O}}(dκ^2ε^{-2})$ and $\tilde{\mathcal{O}}(dε^{-4})$ for NC-SC and NC-C settings, respectively, where $d$ is the dimension and $κ$ is the condition number. We further study the algorithm-dependent generalization bounds via stability arguments of algorithms. In particular, we introduce a novel stability notion for minimax problems and build a connection between generalization bounds and the stability notion. As a result, we establish algorithm-dependent generalization bounds for stochastic gradient descent ascent (SGDA) algorithm and the more general sampling-determined algorithms.

math.OC↗

arXiv2023

Robot Subset Selection for Swarm Lifetime Maximization in Computation Offloading with Correlated Data Sources

Consider robot swarm wireless networks where mobile robots offload their computing tasks to a computing server located at the mobile edge. Our aim is to maximize the swarm lifetime through efficient exploitation of the correlation between distributed data sources. The optimization problem is handled by selecting appropriate robot subsets to send their sensed data to the server. In this work, the data correlation between distributed robot subsets is modelled as an undirected graph. A least-degree iterative partitioning (LDIP) algorithm is proposed to partition the graph into a set of subgraphs. Each subgraph has at least one vertex (i.e., subset), termed representative vertex (R-Vertex), which shares edges with and only with all other vertices within the subgraph; only R-Vertices are selected for data transmissions. When the number of subgraphs is maximized, the proposed subset selection approach is shown to be optimum in the AWGN channel. For independent fading channels, the max-min principle can be incorporated into the proposed approach to achieve the best performance.

eess.SY↗

arXiv2022

A Dynamic 3D Spontaneous Micro-expression Database: Establishment and Evaluation

Micro-expressions are spontaneous, unconscious facial movements that show people's true inner emotions and have great potential in related fields of psychological testing. Since the face is a 3D deformation object, the occurrence of an expression can arouse spatial deformation of the face, but limited by the available databases are 2D videos, lacking the description of 3D spatial information of micro-expressions. Therefore, we proposed a new micro-expression database containing 2D video sequences and 3D point clouds sequences. The database includes 373 micro-expressions sequences, and these samples were classified using the objective method based on facial action coding system, as well as the non-objective method that combines video contents and participants' self-reports. We extracted 2D and 3D features using the local binary patterns on three orthogonal planes (LBP-TOP) and curvature algorithms, respectively, and evaluated the classification accuracies of these two features and their fusion results with leave-one-subject-out (LOSO) and 10-fold cross-validation. Further, we performed various neural network algorithms for database classification, the results show that classification accuracies are improved by fusing 3D features than using only 2D features. The database offers original and cropped micro-expression samples, which will facilitate the exploration and research on 3D Spatio-temporal features of micro-expressions.

cs.CV↗

arXiv2021

Correlation-Based Device Energy-Efficient Dynamic Multi-Task Offloading for Mobile Edge Computing

Task offloading to mobile edge computing (MEC) has emerged as a key technology to alleviate the computation workloads of mobile devices and decrease service latency for the computation-intensive applications. Device battery consumption is one of the limiting factors needs to be considered during task offloading. In this paper, multi-task offloading strategies have been investigated to improve device energy efficiency. Correlations among tasks in time domain as well as task domain are proposed to be employed to reduce the number of tasks to be transmitted to MEC. Furthermore, a binary decision tree based algorithm is investigated to jointly optimize the mobile device clock frequency, transmission power, structure and number of tasks to be transmitted. MATLAB based simulation is employed to demonstrate the performance of our proposed algorithm. It is observed that the proposed dynamic multi-task offloading strategies can reduce the total energy consumption at device along various transmit power versus noise power point compared with the conventional one.

cs.NI↗

arXiv2021

The Complexity of Nonconvex-Strongly-Concave Minimax Optimization

This paper studies the complexity for finding approximate stationary points of nonconvex-strongly-concave (NC-SC) smooth minimax problems, in both general and averaged smooth finite-sum settings. We establish nontrivial lower complexity bounds of $Ω(\sqrtκΔLε^{-2})$ and $Ω(n+\sqrt{nκ}ΔLε^{-2})$ for the two settings, respectively, where $κ$ is the condition number, $L$ is the smoothness constant, and $Δ$ is the initial gap. Our result reveals substantial gaps between these limits and best-known upper bounds in the literature. To close these gaps, we introduce a generic acceleration scheme that deploys existing gradient-based methods to solve a sequence of crafted strongly-convex-strongly-concave subproblems. In the general setting, the complexity of our proposed algorithm nearly matches the lower bound; in particular, it removes an additional poly-logarithmic dependence on accuracy present in previous works. In the averaged smooth finite-sum setting, our proposed algorithm improves over previous algorithms by providing a nearly-tight dependence on the condition number.

math.OC↗

arXiv2020

Scannerless non-line-of-sight three dimensional imaging with a 32x32 SPAD array

We develop a scannerless non-line-of-sight three dimensional imaging system based on a commercial 32x32 SPAD camera combined with a 70 ps pulsed laser. In our experiment, 1024 time histograms can be achieved synchronously in 3s with an average time resolution of about 165 ps. The result with filtered back projection shows a discernable reconstruction while the result using virtual wave field demonstrates a better quality similar to the ones created by earlier scanning imaging systems with single pixel SPAD. Comparatively, our system has large potential advantages in frame frequency, power requirements, compactness and robustness. The research results will pave a path for scannerless non-line-of-sight three dimensional imaging application.

eess.IV↗

arXiv2019

Highly Conductive Co3Se4 Embedded in N-doped 3D Interconnected Carbonaceous Networks for Enhanced Lithium and Sodium Storage

Traditional cobalt selenides as active materials in lithium-ion batteries (LIBs) and sodium-ion batteries (SIBs) would suffer from drastic volume expansions and large stacking effects, leading to a low cycling stability. In this work, we utilized a facile template method for preparing Co3Se4@N-CN (CSNC) that encapsulated Co3Se4 nanoparticles into 3D interconnected nitrogen-doped carbon network (N-CN). Satisfactorily, it possesses excellent cycling stability with enhanced lithium and sodium energy storage capacity. As an anode material in LIBs, CSNC exhibited a prominent reversible discharge performance of 1313.5 mAh g-1 after 100 cycles at 0.1 A g-1 and 835.6 mAh g-1 after 500 cycles at 1.0 A g-1. Interestingly, according to the analysis from cyclic voltammetry, the in-situ generated Se might provide extra capacity that leaded to a rising trend of capacity. When utilized as an anode in SIBs, CSNC delivered an outstanding capacity of 448.7 mAh g-1 after 100 cycles at 0.1 A g-1 and could retain 328.9 mAh g-1 (77.2% of that of 0.1 A g-1) even at a high current density of 5.0 A g-1. The results demonstrate that CSNC is a superior anode material in LIBs and SIBs with great promise. More importantly, this strategy opens up an effective avenue for the design of transition metal selenide/carbonaceous composites for advanced battery storage systems.

physics.app-ph↗

arXiv2018

A New Probability-one Homotopy Method for Solving Minimum-Time Low-Thrust Orbital Transfer Problems

Homotopy methods have been widely utilized to solve low-thrust orbital transfer problems, however, it is not guaranteed that the optimal solution can be obtained by the existing homotopy methods. In this paper, a new homotopy method is presented, by which the optimal solution can be found with probability one. Generalized sufficient conditions, which are derived from the parametrized Sard's theorem, are first developed. A new type of probability-one homotopy formulation, which is custom-designed for solving minimum-time low-thrust trajectory optimization problems and satisfies all these sufficient conditions, is then constructed. By tracking the continuous zero curve initiated by an initial problem with known solution, the optimal solution of the original problem is guaranteed to be solved with probability one. Numerical demonstrations in a three-dimensional time-optimal low-thrust orbital transfer problem with 43 revolutions is presented to illustrate the applications of the method.

astro-ph.IM↗

arXiv2018

On the Convergence Rate of Stochastic Mirror Descent for Nonsmooth Nonconvex Optimization

In this paper, we investigate the non-asymptotic stationary convergence behavior of Stochastic Mirror Descent (SMD) for nonconvex optimization. We focus on a general class of nonconvex nonsmooth stochastic optimization problems, in which the objective can be decomposed into a relatively weakly convex function (possibly non-Lipschitz) and a simple non-smooth convex regularizer. We prove that SMD, without the use of mini-batch, is guaranteed to converge to a stationary point in a convergence rate of $ \mathcal{O}(1/\sqrt{t}) $. The efficiency estimate matches with existing results for stochastic subgradient method, but is evaluated under a stronger stationarity measure. Our convergence analysis applies to both the original SMD and its proximal version, as well as the deterministic variants, for solving relatively weakly convex problems.

math.OC↗