Supersaturated designs, in which the number of factors exceeds the number of runs, are often constructed under a heuristic criterion that measures a design's proximity to an unattainable orthogonal design. Such a criterion does not directly measure a design's quality in terms of screening. To address this disconnect, we develop optimality criteria to maximize the lasso's sign recovery probability. The criteria have varying amounts of prior knowledge about the model's parameters. We show that an orthogonal design is an ideal structure when the signs of the active factors are unknown. When the signs are assumed known, we show that a design whose columns exhibit small, positive correlations are ideal. Such designs are sought after by the Var(s+)-criterion. These conclusions are based on a continuous optimization framework, which rigorously justifies the use of established heuristic criteria. From this justification, we propose a computationally-efficient design search algorithm that filters through optimal designs under different heuristic criteria to select the one that maximizes the sign recovery probability under the lasso.
相關內容
Vehicle teleoperation has numerous potential applications, including serving as a backup solution for autonomous vehicles, facilitating remote delivery services, and enabling hazardous remote operations. However, complex urban scenarios, limited situational awareness, and network delay increase the cognitive workload of human operators and degrade teleoperation performance. To address this, the successive reference pose tracking (SRPT) approach was introduced in earlier work, which transmits successive reference poses to the remote vehicle instead of steering commands. The operator generates reference poses online with the help of a joystick steering and an augmented display, potentially mitigating the detrimental effects of delays. However, it is not clear which minimal set of sensors is essential for the SRPT vehicle teleoperation control loop. This paper tests the robustness of the SRPT approach in the presence of state estimation inaccuracies, environmental disturbances, and measurement noises. The simulation environment, implemented in Simulink, features a 14-dof vehicle model and incorporates difficult maneuvers such as tight corners, double-lane changes, and slalom. Environmental disturbances include low adhesion track regions and strong cross-wind gusts. The results demonstrate that the SRPT approach, using either estimated or actual states, performs similarly under various worst-case scenarios, even without a position sensor requirement. Additionally, the designed state estimator ensures sufficient performance with just an inertial measurement unit, wheel speed encoder, and steer encoder, constituting a minimal set of essential sensors for the SRPT vehicle teleoperation control loop.
Robots with the ability to balance time against the thoroughness of search have the potential to provide time-critical assistance in applications such as search and rescue. Current advances in ergodic coverage-based search methods have enabled robots to completely explore and search an area in a fixed amount of time. However, optimizing time against the quality of autonomous ergodic search has yet to be demonstrated. In this paper, we investigate solutions to the time-optimal ergodic search problem for fast and adaptive robotic search and exploration. We pose the problem as a minimum time problem with an ergodic inequality constraint whose upper bound regulates and balances the granularity of search against time. Solutions to the problem are presented analytically using Pontryagin's conditions of optimality and demonstrated numerically through a direct transcription optimization approach. We show the efficacy of the approach in generating time-optimal ergodic search trajectories in simulation and with drone experiments in a cluttered environment. Obstacle avoidance is shown to be readily integrated into our formulation, and we perform ablation studies that investigate parameter dependence on optimized time and trajectory sensitivity for search.
The novel concept of near-field non-orthogonal multiple access (NF-NOMA) communications is proposed. The near-filed beamfocusing enables NOMA to be carried out in both angular and distance domains. Two novel frameworks are proposed, namely, single-location-beamfocusing NF-NOMA (SLB-NF-NOMA) and multiple-location-beamfocusing NF-NOMA (MLB-NF-NOMA). 1) For SLB-NF-NOMA, two NOMA users in the same angular direction with distinct quality of service (QoS) requirements can be grouped into one cluster. The hybrid beamformer design and power allocation problem is formulated to maximize the sum rate of the users with higher QoS (H-QoS) requirements. To solve this problem, the analog beamformer is first designed to focus the energy on the H-QoS users and the zero-forcing (ZF) digital beamformer is employed. Then, the optimal power allocation is obtained. 2) For MLB-NF-NOMA, the two NOMA users in the same cluster can have different angular directions. The analog beamformer is first designed to focus the energy on both two NOMA users. Then, a singular value decomposition (SVD) based ZF (SVD-ZF) digital beamformer is designed. Furthermore, a novel antenna allocation algorithm is proposed. Finally, a suboptimal power allocation algorithm is proposed. Numerical results demonstrate that the NF-NOMA can achieve a higher spectral efficiency and provide a higher flexibility than conventional far-field NOMA.
We give query complexity lower bounds for convex optimization and the related feasibility problem. We show that quadratic memory is necessary to achieve the optimal oracle complexity for first-order convex optimization. In particular, this shows that center-of-mass cutting-planes algorithms in dimension $d$ which use $\tilde O(d^2)$ memory and $\tilde O(d)$ queries are Pareto-optimal for both convex optimization and the feasibility problem, up to logarithmic factors. Precisely, we prove that to minimize $1$-Lipschitz convex functions over the unit ball to $1/d^4$ accuracy, any deterministic first-order algorithms using at most $d^{2-\delta}$ bits of memory must make $\tilde\Omega(d^{1+\delta/3})$ queries, for any $\delta\in[0,1]$. For the feasibility problem, in which an algorithm only has access to a separation oracle, we show a stronger trade-off: for at most $d^{2-\delta}$ memory, the number of queries required is $\tilde\Omega(d^{1+\delta})$. This resolves a COLT 2019 open problem of Woodworth and Srebro.
For solving linear inverse problems, particularly of the type that appear in tomographic imaging and compressive sensing, this paper develops two new approaches. The first approach is an iterative algorithm that minimizers a regularized least squares objective function where the regularization is based on a compound Gaussian prior distribution. The Compound Gaussian prior subsumes many of the commonly used priors in image reconstruction, including those of sparsity-based approaches. The developed iterative algorithm gives rise to the paper's second new approach, which is a deep neural network that corresponds to an "unrolling" or "unfolding" of the iterative algorithm. Unrolled deep neural networks have interpretable layers and outperform standard deep learning methods. This paper includes a detailed computational theory that provides insight into the construction and performance of both algorithms. The conclusion is that both algorithms outperform other state-of-the-art approaches to tomographic image formation and compressive sensing, especially in the difficult regime of low training.
Computer graphics images (CGIs) are artificially generated by means of computer programs and are widely perceived under various scenarios, such as games, streaming media, etc. In practice, the quality of CGIs consistently suffers from poor rendering during production, inevitable compression artifacts during the transmission of multimedia applications, and low aesthetic quality resulting from poor composition and design. However, few works have been dedicated to dealing with the challenge of computer graphics image quality assessment (CGIQA). Most image quality assessment (IQA) metrics are developed for natural scene images (NSIs) and validated on databases consisting of NSIs with synthetic distortions, which are not suitable for in-the-wild CGIs. To bridge the gap between evaluating the quality of NSIs and CGIs, we construct a large-scale in-the-wild CGIQA database consisting of 6,000 CGIs (CGIQA-6k) and carry out the subjective experiment in a well-controlled laboratory environment to obtain the accurate perceptual ratings of the CGIs. Then, we propose an effective deep learning-based no-reference (NR) IQA model by utilizing both distortion and aesthetic quality representation. Experimental results show that the proposed method outperforms all other state-of-the-art NR IQA methods on the constructed CGIQA-6k database and other CGIQA-related databases. The database will be released to facilitate further research.
We study the 2D Navier-Stokes equation with transport noise subject to periodic boundary conditions. Our main result is an error estimate for the time-discretisation showing a convergence rate of order (up to) 1/2. It holds with respect to mean square error convergence, whereas previously such a rate for the stochastic Navier-Stokes equations was only known with respect to convergence in probability. Our result is based on uniform-in-probability estimates for the continuous as well as the time-discrete solution exploiting the particular structure of the noise.
The random forest (RF) algorithm has become a very popular prediction method for its great flexibility and promising accuracy. In RF, it is conventional to put equal weights on all the base learners (trees) to aggregate their predictions. However, the predictive performances of different trees within the forest can be very different due to the randomization of the embedded bootstrap sampling and feature selection. In this paper, we focus on RF for regression and propose two optimal weighting algorithms, namely the 1 Step Optimal Weighted RF (1step-WRF$_\mathrm{opt}$) and 2 Steps Optimal Weighted RF (2steps-WRF$_\mathrm{opt}$), that combine the base learners through the weights determined by weight choice criteria. Under some regularity conditions, we show that these algorithms are asymptotically optimal in the sense that the resulting squared loss and risk are asymptotically identical to those of the infeasible but best possible model averaging estimator. Numerical studies conducted on real-world data sets indicate that these algorithms outperform the equal-weight forest and two other weighted RFs proposed in existing literature in most cases.
Employees work in increasingly digital environments that enable advanced analytics. Yet, they lack oversight over the systems that process their data. That means that potential analysis errors or hidden biases are hard to uncover. Recent data protection legislation tries to tackle these issues, but it is inadequate. It does not prevent data misusage while at the same time stifling sensible use cases for data. We think the conflict between data protection and increasingly data-driven systems should be solved differently. When access to an employees' data is given, all usages should be made transparent to them, according to the concept of inverse transparency. This allows individuals to benefit from sensible data usage while addressing the potentially harmful consequences of data misusage. To accomplish this, we propose a new design approach for workforce analytics we refer to as inverse transparency by design. To understand the developer and user perspectives on the proposal, we conduct two exploratory studies with students. First, we let small teams of developers implement analytics tools with inverse transparency by design to uncover how they judge the approach and how it materializes in their developed tools. We find that architectural changes are made without inhibiting core functionality. The developers consider our approach valuable and technically feasible. Second, we conduct a user study over three months to let participants experience the provided inverse transparency and reflect on their experience. The study models a software development workplace where most work processes are already digital. Participants perceive the transparency as beneficial and feel empowered by it. They unanimously agree that it would be an improvement for the workplace. We conclude that inverse transparency by design is a promising approach to realize accepted and responsible people analytics.
Image-to-image translation aims to learn the mapping between two visual domains. There are two main challenges for many applications: 1) the lack of aligned training pairs and 2) multiple possible outputs from a single input image. In this work, we present an approach based on disentangled representation for producing diverse outputs without paired training images. To achieve diversity, we propose to embed images onto two spaces: a domain-invariant content space capturing shared information across domains and a domain-specific attribute space. Our model takes the encoded content features extracted from a given input and the attribute vectors sampled from the attribute space to produce diverse outputs at test time. To handle unpaired training data, we introduce a novel cross-cycle consistency loss based on disentangled representations. Qualitative results show that our model can generate diverse and realistic images on a wide range of tasks without paired training data. For quantitative comparisons, we measure realism with user study and diversity with a perceptual distance metric. We apply the proposed model to domain adaptation and show competitive performance when compared to the state-of-the-art on the MNIST-M and the LineMod datasets.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191