We tackle the problem of efficiently approximating the volume of convex polytopes, when these are given in three different representations: H-polytopes, which have been studied extensively, V-polytopes, and zonotopes (Z-polytopes). We design a novel practical Multiphase Monte Carlo algorithm that leverages random walks based on billiard trajectories, as well as a new empirical convergence tests and a simulated annealing schedule of adaptive convex bodies. After tuning several parameters of our proposed method, we present a detailed experimental evaluation of our tuned algorithm using a rich dataset containing Birkhoff polytopes and polytopes from structural biology. Our open-source implementation tackles problems that have been intractable so far, offering the first software to scale up in thousands of dimensions for H-polytopes and in the hundreds for V- and Z-polytopes on moderate hardware. Last, we illustrate our software in evaluating Z-polytope approximations.
相關內容
Offline estimation of the dynamical model of a Markov Decision Process (MDP) is a non-trivial task that greatly depends on the data available in the learning phase. Sometimes the dynamics of the model is invariant with respect to some transformations of the current state and action. Recent works showed that an expert-guided pipeline relying on Density Estimation methods as Deep Neural Network based Normalizing Flows effectively detects this structure in deterministic environments, both categorical and continuous-valued. The acquired knowledge can be exploited to augment the original data set, leading eventually to a reduction in the distributional shift between the true and the learned model. Such data augmentation technique can be exploited as a preliminary process to be executed before adopting an Offline Reinforcement Learning architecture, increasing its performance. In this work we extend the paradigm to also tackle non-deterministic MDPs, in particular, 1) we propose a detection threshold in categorical environments based on statistical distances, and 2) we show that the former results lead to a performance improvement when solving the learned MDP and then applying the optimized policy in the real environment.
We present a deep learning method for accurately localizing the center of a single corneal reflection (CR) in an eye image. Unlike previous approaches, we use a convolutional neural network (CNN) that was trained solely using simulated data. Using only simulated data has the benefit of completely sidestepping the time-consuming process of manual annotation that is required for supervised training on real eye images. To systematically evaluate the accuracy of our method, we first tested it on images with simulated CRs placed on different backgrounds and embedded in varying levels of noise. Second, we tested the method on high-quality videos captured from real eyes. Our method outperformed state-of-the-art algorithmic methods on real eye images with a 35% reduction in terms of spatial precision, and performed on par with state-of-the-art on simulated images in terms of spatial accuracy.We conclude that our method provides a precise method for CR center localization and provides a solution to the data availability problem which is one of the important common roadblocks in the development of deep learning models for gaze estimation. Due to the superior CR center localization and ease of application, our method has the potential to improve the accuracy and precision of CR-based eye trackers
Learning image classification and image generation using the same set of network parameters is a challenging problem. Recent advanced approaches perform well in one task often exhibit poor performance in the other. This work introduces an energy-based classifier and generator, namely EGC, which can achieve superior performance in both tasks using a single neural network. Unlike a conventional classifier that outputs a label given an image (i.e., a conditional distribution $p(y|\mathbf{x})$), the forward pass in EGC is a classifier that outputs a joint distribution $p(\mathbf{x},y)$, enabling an image generator in its backward pass by marginalizing out the label $y$. This is done by estimating the energy and classification probability given a noisy image in the forward pass, while denoising it using the score function estimated in the backward pass. EGC achieves competitive generation results compared with state-of-the-art approaches on ImageNet-1k, CelebA-HQ and LSUN Church, while achieving superior classification accuracy and robustness against adversarial attacks on CIFAR-10. This work represents the first successful attempt to simultaneously excel in both tasks using a single set of network parameters. We believe that EGC bridges the gap between discriminative and generative learning.
Aerial scans with unmanned aerial vehicles (UAVs) are becoming more widely adopted across industries, from smart farming to urban mapping. An application area that can leverage the strength of such systems is search and rescue (SAR) operations. However, with a vast variability in strategies and topology of application scenarios, as well as the difficulties in setting up real-world UAV-aided SAR operations for testing, designing an optimal flight pattern to search for and detect all victims can be a challenging problem. Specifically, the deployed UAV should be able to scan the area in the shortest amount of time while maintaining high victim detection recall rates. Therefore, low probability of false negatives (i.e., high recall) is more important than precision in this case. To address the issues mentioned above, we have developed a simulation environment that emulates different SAR scenarios and allows experimentation with flight missions to provide insight into their efficiency. The solution was developed with the open-source ROS framework and Gazebo simulator, with PX4 as the autopilot system for flight control, and YOLO as the object detector.
Offline reinforcement learning (RL) have received rising interest due to its appealing data efficiency. The present study addresses behavior estimation, a task that lays the foundation of many offline RL algorithms. Behavior estimation aims at estimating the policy with which training data are generated. In particular, this work considers a scenario where the data are collected from multiple sources. In this case, neglecting data heterogeneity, existing approaches for behavior estimation suffers from behavior misspecification. To overcome this drawback, the present study proposes a latent variable model to infer a set of policies from data, which allows an agent to use as behavior policy the policy that best describes a particular trajectory. This model provides with a agent fine-grained characterization for multi-source data and helps it overcome behavior misspecification. This work also proposes a learning algorithm for this model and illustrates its practical usage via extending an existing offline RL algorithm. Lastly, with extensive evaluation this work confirms the existence of behavior misspecification and the efficacy of the proposed model.
Iterative minimization algorithms appear in various areas including machine learning, neural network, and information theory. The em algorithm is one of the famous one in the former area, and Arimoto-Blahut algorithm is a typical one in the latter area. However, these two topics had been separately studied for a long time. In this paper, we generalize an algorithm that was recently proposed in the context of Arimoto-Blahut algorithm. Then, we show various convergence theorems, one of which covers the case when each iterative step is done approximately. Also, we apply this algorithm to the target problem in em algorithm, and propose its improvement. In addition, we apply it to other various problems in information theory.
Trajectory optimization problems for legged robots are commonly formulated with fixed contact schedules. These multi-phase Hybrid Trajectory Optimization (HTO) methods result in locally optimal trajectories, but the result depends heavily upon the predefined contact mode sequence. Contact-Implicit Optimization (CIO) offers a potential solution to this issue by allowing the contact mode to be determined throughout the trajectory by the optimization solver. However, CIO suffers from long solve times and convergence issues. This work combines the benefits of these two methods into one algorithm: Staged Contact Optimization (SCO). SCO tightens constraints on contact in stages, eventually fixing them to allow robust and fast convergence to a feasible solution. Results on a planar biped and spatial quadruped demonstrate speed and optimality improvements over CIO and HTO. These properties make SCO well suited for offline trajectory generation or as an effective tool for exploring the dynamic capabilities of a robot.
Adhesive joints are increasingly used in industry for a wide variety of applications because of their favorable characteristics such as high strength-to-weight ratio, design flexibility, limited stress concentrations, planar force transfer, good damage tolerance, and fatigue resistance. Finding the optimal process parameters for an adhesive bonding process is challenging: the optimization is inherently multi-objective (aiming to maximize break strength while minimizing cost), constrained (the process should not result in any visual damage to the materials, and stress tests should not result in failures that are adhesion-related), and uncertain (testing the same process parameters several times may lead to different break strengths). Real-life physical experiments in the lab are expensive to perform. Traditional evolutionary approaches (such as genetic algorithms) are then ill-suited to solve the problem, due to the prohibitive amount of experiments required for evaluation. Although Bayesian optimization-based algorithms are preferred to solve such expensive problems, few methods consider the optimization of more than one (noisy) objective and several constraints at the same time. In this research, we successfully applied specific machine learning techniques (Gaussian Process Regression) to emulate the objective and constraint functions based on a limited amount of experimental data. The techniques are embedded in a Bayesian optimization algorithm, which succeeds in detecting Pareto-optimal process settings in a highly efficient way (i.e., requiring a limited number of physical experiments).
Hashing is very popular for remote sensing image search. This article proposes a multiview hashing with learnable parameters to retrieve the queried images for a large-scale remote sensing dataset. Existing methods always neglect that real-world remote sensing data lies on a low-dimensional manifold embedded in high-dimensional ambient space. Unlike previous methods, this article proposes to learn the consensus compact codes in a view-specific low-dimensional subspace. Furthermore, we have added a hyperparameter learnable module to avoid complex parameter tuning. In order to prove the effectiveness of our method, we carried out experiments on three widely used remote sensing data sets and compared them with seven state-of-the-art methods. Extensive experiments show that the proposed method can achieve competitive results compared to the other method.
The indeterminate nature of human motion requires trajectory prediction systems to use a probabilistic model to formulate the multi-modality phenomenon and infer a finite set of future trajectories. However, the inference processes of most existing methods rely on Monte Carlo random sampling, which is insufficient to cover the realistic paths with finite samples, due to the long tail effect of the predicted distribution. To promote the sampling process of stochastic prediction, we propose a novel method, called BOsampler, to adaptively mine potential paths with Bayesian optimization in an unsupervised manner, as a sequential design strategy in which new prediction is dependent on the previously drawn samples. Specifically, we model the trajectory sampling as a Gaussian process and construct an acquisition function to measure the potential sampling value. This acquisition function applies the original distribution as prior and encourages exploring paths in the long-tail region. This sampling method can be integrated with existing stochastic predictive models without retraining. Experimental results on various baseline methods demonstrate the effectiveness of our method.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191