Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Incremental Structure-from-Motion is a prevalent strategy for 3D reconstruction from unordered image collections. While incremental reconstruction systems have tremendously advanced in all regards, robustness, accuracy, completeness, and scalability remain the key problems towards building a truly general-purpose pipeline. We propose a new SfM technique that improves upon the state of the art to make a further step towards this ultimate goal. The full reconstruction pipeline is released to the public as an open-source implementation.

/pdf/structure-from-motion-revisited-10382wusrz.pdf

Structure-from-Motion Revisited

On the truck dispatching problem

At some point in the future, how far out we do not exactly know, wireless access to the Internet will outstrip all other forms of access bringing the freedom of mobility to the way we access the we...

ACM SIGCOMM computer communication review

Transportation research. part c, emerging technologies

An active robot system can change its visual parameters in an intentional manner and perform its sensing actions purposefully. A general vision task thus can be performed in an efficient way by means of strategic control of the perception process. The controllable processes include 3D active sensing, sensor configuration and recalibration, automatic sensor placement, and 3D sensing. This book explores these important issues in studying for active visual perception. Vision sensors have limited fields of views and can only "see" a portion of a scene from a single viewpoint. To make the entire object visible, the sensor has to be moved from one place to another around the object to observe all features of interest. The sensor planning presented in this book describes an effective strategy to generate a sequence of viewing poses and sensor settings for optimally completing a perception task. Several methods are proposed to solve the problems in both model-based and nonmodel-based vision tasks. For model-based applications, the method involves determination of the optimal sensor placements and a shortest path through these viewpoints for automatic generation of a perception plan. A topology of viewpoints is achieved by a genetic algorithm in which a min-max criterion is used for evaluation. A shortest path is also determined by graph algorithms. For nonmodel-based applications, the method involves determination of the best next view and sensor settings. The trend surface is proposed as the cue to predict the unknown portion of an object or environment. The 11 chapters in Active Vision Planning draw on recent work in robot vision over ten years, particularly in the use of new concepts of active sensing, reconfiguration, recalibration, sensor model, sensing constraints, sensing evaluation, viewpoint decision, sensor placement graph, model based planning, path planning, planning for robot in unknown environment, dynamic 3D construction, surface prediction, etc. Implementation examples are also provided with theoretical methods for testing in a real robot system. With these optimal sensor planning strategies, this book will give the robot vision system the adaptability needed in many practical applications.

Active Sensor Planning for Multiview Vision Tasks

Multi-objective particle swarm optimization based on global margin ranking

In this paper, we consider the robust face recognition problem via iterative re-constrained group sparse classifier (IRGSC) with adaptive weights learning. Specifically, we propose a group sparse representation classification (GSRC) approach in which weighted features and groups are collaboratively adopted to encode more structure information and discriminative information than other regression based methods. In addition, we derive an efficient algorithm to optimize the proposed objective function, and theoretically prove the convergence. There are several appealing aspects associated with IRGSC. First, adaptively learned weights can be seamlessly incorporated into the GSRC framework. This integrates the locality structure of the data and validity information of the features into $l_{2,p}$ -norm regularization to form a unified formulation. Second, IRGSC is very flexible to different size of training set as well as feature dimension thanks to the $l_{2,p}$ -norm regularization. Third, the derived solution is proved to be a stationary point (globally optimal if $p \geq 1$ ). Comprehensive experiments on representative data sets demonstrate that IRGSC is a robust discriminative classifier which significantly improves the performance and efficiency compared with the state-of-the-art methods in dealing with face occlusion, corruption, and illumination changes, and so on.

Iterative Re-Constrained Group Sparse Face Recognition With Adaptive Weights Learning

This article summarizes some methods from biological intelligence for modeling and optimization of supply chain management (SCM) systems, including genetic algorithms, evolutionary programming, differential evolution, swarm intelligence, artificial immune, and other biological intelligence related methods. An SCM system is adaptive, dynamic, open self-organizing, which is maintained by flows of information, materials, goods, funds, and energy. Traditional methods for modeling and optimizing complex SCM systems require huge amounts of computing resources, and biological intelligence-based solutions can often provide valuable alternatives for efficiently solving problems. The paper summarizes the recent related methods for the design and optimization of SCM systems, which covers the most widely used genetic algorithms and other evolutionary algorithms.

/pdf/modeling-of-biological-intelligence-for-scm-system-21pd40womk.pdf

Modeling of Biological Intelligence for SCM System Optimization

Sustainable energy development always involves complex optimization problems of design, planning, and control, which are often computationally difficult for conventional optimization methods. Fortunately, the continuous advances in artificial intelligence have resulted in an increasing number of heuristic optimization methods for effectively handling those complicated problems. Particularly, algorithms that are inspired by the principles of natural biological evolution and/or collective behavior of social colonies have shown a promising performance and are becoming more and more popular nowadays. In this paper we summarize the recent advances in bio-inspired optimization methods, including artificial neural networks, evolutionary algorithms, swarm intelligence, and their hybridizations, which are applied to the field of sustainable energy development. Literature reviewed in this paper shows the current state of the art and discusses the potential future research trends.

/pdf/bio-inspired-optimization-of-sustainable-energy-systems-a-2x4fj1loay.pdf

Wanliang Wang

Papers

Active Sensor Planning for Multiview Vision Tasks

Multi-objective particle swarm optimization based on global margin ranking

Iterative Re-Constrained Group Sparse Face Recognition With Adaptive Weights Learning

Modeling of Biological Intelligence for SCM System Optimization

Bio-Inspired Optimization of Sustainable Energy Systems: A Review