Abstract Triangulation refers to the problem of finding a 3D point from its 2D projections on multiple images taken at different camera poses. For solving this problem, it is a common practice to use the so‐called optimal triangulation method. But, the method can be optimal only if we assume no uncertainty in the camera parameters. In this paper, an extensive comparison between various existing methods for the calibrated cameras is performed, in which different poses are considered. Furthermore, uncertainty sensitivity analysis is conducted for extrinsic parameters of cameras. The results show that the optimal triangulation method is actually not the best choice when there is uncertainty in extrinsic parameters. Interestingly, it can be observed that the simple midpoint method works equally well or outperforms the optimal triangulation and many other methods. Apart from its high performance, the midpoint method has a simple closed form solution for multiple views while the optimal triangulation method is hard to be used for more than two views. Therefore, in contrast to the common practice, we argue that the simple midpoint method can be a good choice in the structure‐from‐motion process where there is uncertainty in extrinsic camera parameters.

This work concentrates on optimization on Riemannian manifolds. The Limited-memory Broyden-Fletcher-Goldfarb-Shanno (LBFGS) algorithm is a commonly used quasi-Newton method for numerical optimization in Euclidean spaces. Riemannian LBFGS (RLBFGS) is an extension of this method to Riemannian manifolds. RLBFGS involves computationally expensive vector transports as well as unfolding recursions using adjoint vector transports. In this article, we propose two mappings in the tangent space using the inverse second root and Cholesky decomposition. These mappings make both vector transport and adjoint vector transport identity and therefore isometric. Identity vector transport makes RLBFGS less computationally expensive and its isometry is also very useful in convergence analysis of RLBFGS. Moreover, under the proposed mappings, the Riemannian metric reduces to Euclidean inner product, which is much less computationally expensive. We focus on the Symmetric Positive Definite (SPD) manifolds which are beneficial in various fields such as data science and statistics. This work opens a research opportunity for extension of the proposed mappings to other well-known manifolds.

Pose graph optimization (PGO), or equivalently pose synchronization, that is synchronizing rotations and positions, is the state-of-the-art formulation for simultaneous localization and mapping in robotics. In this article, we first present a new manifold Gauss-Newton method for solving the rotation synchronization problem. In this article, we derive an efficient implementation of the method and develop a convergence theory thereof. A structural parameter appears in the proof, which has a significant influence on the convergence basin. In this article, we show that this structural parameter is the norm of the inverse of the reduced graph Laplacian and obtains the explicit relation of this parameter for special graph structures. We also present another method that directly solves the pose synchronization using both relative rotation and translation observations. Experimental results show that our rotation synchronization method can be successfully used to initialize iterative PGO solvers. Furthermore, we show that our pose synchronization algorithm outperforms state-of-the-art solvers in high-noise cases.

Learning methods are challenged when there is not enough labelled data. It gets worse when the existing learning data have different distributions in different domains. To deal with such situations, deep unsupervised domain adaptation techniques have newly been widely used. This study surveys such domain adaptation methods that have been used for classification tasks in computer vision. The survey includes the very recent papers on this topic that have not been included in the previous surveys and introduces a taxonomy by grouping methods published on unsupervised domain adaptation into five groups of discrepancy‐, adversarial‐, reconstruction‐, representation‐, and attention‐based methods.

Due to the lack of enough generalization in the state space, common methods of reinforcement learning suffer from slow learning speed, especially in the early learning trials. This paper introduces a model-based method in discrete state spaces for increasing the learning speed in terms of required experiences (but not required computation time) by exploiting generalization in the experiences of the subspaces. A subspace is formed by choosing a subset of features in the original state representation. Generalization and faster learning in a subspace are due to many-to-one mapping of experiences from the state space to each state in the subspace. Nevertheless, due to inherent perceptual aliasing (PA) in the subspaces, the policy suggested by each subspace does not generally converge to the optimal policy. Our approach, called model-based learning with subspaces (MoBLeSs), calculates the confidence intervals of the estimated Q -values in the state space and in the subspaces. These confidence intervals are used in the decision-making, such that the agent benefits the most from the possible generalization while avoiding from the detriment of the PA in the subspaces. The convergence of MoBLeS to the optimal policy is theoretically investigated. In addition, we show through several experiments that MoBLeS improves the learning speed in the early trials.

Pose Graph Optimization (PGO) is an important optimization problem arising in robotics and machine vision applications like 3D reconstruction and 3D SLAM. Each node of pose graph corresponds to an orientation and a location. The PGO problem finds orientations and locations of the nodes from relative noisy observation between nodes. Recent investigations show that well-known iterative PGO solvers need good initialization to converge to good solutions. However, we observed that state-of-the-art initialization methods obtain good initialization only in low noise problems, and they fail in challenging problems having more measurement noise. Consequently, iterative methods may converge to bad solutions in high noise problems. In this paper, a new method for obtaining orientations in the PGO optimization problem is presented. Like other well-known methods the initial locations are obtained from the result of a least-squares problem. The proposed method iteratively approximates the problem around current estimation and converts it to a least-squares problem. Therefore, the method can be seen as an iterative least-squares method which is computationally efficient. Simulation results show that the proposed initialization method helps the most well-known iterative solver to obtain better optima and significantly outperform other solvers in some cases.

We revisit the task of learning a Euclidean metric from data. We approach this problem from first principles and formulate it as a surprisingly simple optimization problem. Indeed, our formulation even admits a closed form solution. This solution possesses several very attractive properties: (i) an innate geometric appeal through the Riemannian geometry of positive definite matrices; (ii) ease of interpretability; and (iii) computational speed several orders of magnitude faster than the widely used LMNN and ITML methods. Furthermore, on standard benchmark datasets, our closed-form solution consistently attains higher classification accuracy.

We study mixture modeling using the elliptical gamma (EG) distribution, a non-Gaussian distribution that allows heavy and light tail and peak behaviors. We first consider maximum likelihood parameter estimation, a task that turns out to be very challenging: we must handle positive definiteness constraints, and more crucially, we must handle possibly nonconcave log-likelihoods, which makes maximization hard. We overcome these difficulties by developing algorithms based on fixed-point theory; our methods respect the psd constraint, while also efficiently solving the (possibly) nonconcave maximization to global optimality. Subsequently, we focus on mixture modeling using EG distributions: we present a closed-form expression of the KL-divergence between two EG distributions, which we then combine with our ML estimation methods to obtain an efficient split-and-merge expectation maximization algorithm. We illustrate the use of our model and algorithms on a dataset of natural image patches.

We take a new look at parameter estimation for Gaussian Mixture Model (GMMs). Specifically, we advance Riemannian manifold optimization (on the manifold of positive definite matrices) as a potential replacement for Expectation Maximization (EM), which has been the de facto standard for decades. An out-of-the-box invocation of Riemannian optimization, however, fails spectacularly: it obtains the same solution as EM, but vastly slower. Building on intuition from geometric convexity, we propose a simple reformulation that has remarkable consequences: it makes Riemannian optimization not only match EM (a nontrivial result on its own, given the poor record nonlinear programming has had against EM), but also outperform it in many settings. To bring our ideas to fruition, we develop a well-tuned Riemannian LBFGS method that proves superior to known competing methods (e.g., Riemannian conjugate gradient). We hope that our results encourage a wider consideration of manifold optimization in machine learning and statistics.

Hermitian positive definite matrices (HPD) recur throughout statistics and machine learning. In this paper we develop \emph{geometric optimisation} for globally optimising certain nonconvex loss functions arising in the modelling of data via elliptically contoured distributions (ECDs). We exploit the remarkable structure of the convex cone of positive definite matrices which allows one to uncover hidden geodesic convexity of objective functions that are nonconvex in the ordinary Euclidean sense. Going even beyond manifold convexity we show how further metric properties of HPD matrices can be exploited to globally optimise several ECD log-likelihoods that are not even geodesic convex. We present key results that help recognise this geometric structure, as well as obtain efficient fixed-point algorithms to optimise the corresponding objective functions. To our knowledge, ours are the most general results on geometric optimisation of HPD matrices known so far. Experiments reveal the benefits of our approach---it avoids any eigenvalue computations which makes it very competitive.