Author Archives: Juan-antonio Fernández-madrigal

Brief but nice related work about structured prediction (MRFs, CRFs, etc.)

Bratieres, S.; Quadrianto, N.; Ghahramani, Z., GPstruct: Bayesian Structured Prediction Using Gaussian Processes, Pattern Analysis and Machine Intelligence, IEEE Transactions on , vol.37, no.7, pp.1514,1520, July 1 2015, DOI: 10.1109/TPAMI.2014.2366151.

We introduce a conceptually novel structured prediction model, GPstruct, which is kernelized, non-parametric and Bayesian, by design. We motivate the model with respect to existing approaches, among others, conditional random fields (CRFs), maximum margin Markov networks (M ^3 N), and structured support vector machines (SVMstruct), which embody only a subset of its properties. We present an inference procedure based on Markov Chain Monte Carlo. The framework can be instantiated for a wide range of structured objects such as linear chains, trees, grids, and other general graphs. As a proof of concept, the model is benchmarked on several natural language processing tasks and a video gesture segmentation task involving a linear chain structure. We show prediction accuracies for GPstruct which are comparable to or exceeding those of CRFs and SVMstruct.

Accelerating the updating stage of a PF through selection of a few representative particles and interpolation of their weights to the rest, with interesting methods for selection and interpolation and a nice related work of efficiency-improved PFs

Shabat, G.; Shmueli, Y.; Bermanis, A.; Averbuch, A., Accelerating Particle Filter Using Randomized Multiscale and Fast Multipole Type Methods, Pattern Analysis and Machine Intelligence, IEEE Transactions on , vol.37, no.7, pp.1396,1407, July 1 2015, DOI: 10.1109/TPAMI.2015.2392754.

Particle filter is a powerful tool for state tracking using non-linear observations. We present a multiscale based method that accelerates the tracking computation by particle filters. Unlike the conventional way, which calculates weights over all particles in each cycle of the algorithm, we sample a small subset from the source particles using matrix decomposition methods. Then, we apply a function extension algorithm that uses a particle subset to recover the density function for all the rest of the particles not included in the chosen subset. The computational effort is substantial especially when multiple objects are tracked concurrently. The proposed algorithm significantly reduces the computational load. By using the Fast Gaussian Transform, the complexity of the particle selection step is reduced to a linear time in n and k , where n is the number of particles and k is the number of particles in the selected subset. We demonstrate our method on both simulated and on real data such as object tracking in video sequences.

Deducing the space concept from the sensorimotor behaviour of a robot, and an interesting related work of uninterpreted sensors and actuators in developmental robotics that deserves a deeper look

Alban Laflaquière, J. Kevin O’Regan, Sylvain Argentieri, Bruno Gas, Alexander V. Terekhov, Learning agent’s spatial configuration from sensorimotor invariants, Robotics and Autonomous Systems, Volume 71, September 2015, Pages 49-59, ISSN 0921-8890, DOI: 10.1016/j.robot.2015.01.003.

The design of robotic systems is largely dictated by our purely human intuition about how we perceive the world. This intuition has been proven incorrect with regard to a number of critical issues, such as visual change blindness. In order to develop truly autonomous robots, we must step away from this intuition and let robotic agents develop their own way of perceiving. The robot should start from scratch and gradually develop perceptual notions, under no prior assumptions, exclusively by looking into its sensorimotor experience and identifying repetitive patterns and invariants. One of the most fundamental perceptual notions, space, cannot be an exception to this requirement. In this paper we look into the prerequisites for the emergence of simplified spatial notions on the basis of a robot’s sensorimotor flow. We show that the notion of space as environment-independent cannot be deduced solely from exteroceptive information, which is highly variable and is mainly determined by the contents of the environment. The environment-independent definition of space can be approached by looking into the functions that link the motor commands to changes in exteroceptive inputs. In a sufficiently rich environment, the kernels of these functions correspond uniquely to the spatial configuration of the agent’s exteroceptors. We simulate a redundant robotic arm with a retina installed at its end-point and show how this agent can learn the configuration space of its retina. The resulting manifold has the topology of the Cartesian product of a plane and a circle, and corresponds to the planar position and orientation of the retina.

A new approach to solve POMDP-like problems through gradient descent and optimal control

Vadim Indelman, Luca Carlone, Frank Dellaert, Planning in the continuous domain: A generalized belief space approach for autonomous navigation in unknown environments, The International Journal of Robotics Research, vol. 34 no. 7, pp. 849-882, DOI: 10.1177/0278364914561102.

We investigate the problem of planning under uncertainty, with application to mobile robotics. We propose a probabilistic framework in which the robot bases its decisions on the generalized belief, which is a probabilistic description of its own state and of external variables of interest. The approach naturally leads to a dual-layer architecture: an inner estimation layer, which performs inference to predict the outcome of possible decisions; and an outer decisional layer which is in charge of deciding the best action to undertake. Decision making is entrusted to a model predictive control (MPC) scheme. The formulation is valid for general cost functions and does not discretize the state or control space, enabling planning in continuous domain. Moreover, it allows to relax the assumption of maximum likelihood observations: predicted measurements are treated as random variables, and binary random variables are used to model the event that a measurement is actually taken by the robot. We successfully apply our approach to the problem of uncertainty-constrained exploration, in which the robot has to perform tasks in an unknown environment, while maintaining localization uncertainty within given bounds. We present an extensive numerical analysis of the proposed approach and compare it against related work. In practice, our planning approach produces smooth and natural trajectories and is able to impose soft upper bounds on the uncertainty. Finally, we exploit the results of this analysis to identify current limitations and show that the proposed framework can accommodate several desirable extensions.

Neural support for the cognitive map: place cells and grid cells

Kate J. Jeffery, Distorting the metric fabric of the cognitive map, Trends in Cognitive Sciences, Volume 19, Issue 6, June 2015, Pages 300-301, ISSN 1364-6613, DOI: 10.1016/j.tics.2015.04.001..

Grid cells are neurons whose regularly spaced firing fields form apparently symmetric arrays, or grids, that are thought to collectively provide an environment-independent metric framework for the brain’s cognitive map of space. However, two recent studies show that grids are naturally distorted, revealing greater local environment-specific effects than previously recognized.

A quick, formal explanation of the PageRank algorithm and its existing variants

Lei, J.; Chen, H., Distributed Randomized PageRank Algorithm Based on Stochastic Approximation, Automatic Control, IEEE Transactions on , vol.60, no.6, pp.1641,1646, June 2015. DOI: 10.1109/TAC.2014.2359311.

A distributed randomized PageRank algorithm based on stochastic approximation (SA) is proposed to estimate the importance scores of web pages. Compared with the existing methods, the algorithm given here has wider applications in the sense that it can deal with a larger class of randomizations. The strong consistency of the estimates is proved, and the robustness of the PageRank value is analyzed as well. Numerical examples are given to verify the obtained theoretic results.

Soft-real time scheduling in distributed systems based on accrued utility of distributable threads, including situations of node failures

Ravindran B., Anderson J. S., Jensen E. D., On Distributed Real-Time Scheduling in Networked Embedded Systems in the Presence of Crash Failures, Lecture Notes in Computer Science, vol. 4761, pp. 67-81, DOI: 10.1007/978-3-540-75664-4_8.

We consider the problem of scheduling distributable real-time threads in networkedembedded systems that operate under run-time uncertainties including those on thread execution times, thread arrivals, and node failure occurrences. We present a distributed scheduling algorithm called CUA. We show that CUA satisfies thread time constraints in the presence of crash failures, is early-deciding, has an efficient message complexity of O(f n) (where f is the number of crashes that actually occur and n is the number of nodes), and is time-optimal with a time lower bound of O(D + f d + nk) (where D is the message delay upper bound, d is the failure detection bound, and k is the maximum number of threads). In crash-free runs, the algorithm constructs schedules within O(D + nk), and yields optimal total utility if nodes are also not overloaded. The algorithm is also “best-effort” in that a high importance thread that may arrive at any time has a very high likelihood for feasible completion (in contrast to classical admission control algorithms which favor feasible completion of admitted threads over admitting new ones, irrespective of thread importance).

A framework to manage and switch between several sensor modalities in tele-operation

Andrea Cherubini, Robin Passama, Philippe Fraisse, André Crosnier, A unified multimodal control framework for human–robot interaction, Robotics and Autonomous Systems, Volume 70, August 2015, Pages 106-115, ISSN 0921-8890, DOI: 10.1016/j.robot.2015.03.002.

In human–robot interaction, the robot controller must reactively adapt to sudden changes in the environment (due to unpredictable human behaviour). This often requires operating different modes, and managing sudden signal changes from heterogeneous sensor data. In this paper, we present a multimodal sensor-based controller, enabling a robot to adapt to changes in the sensor signals (here, changes in the human collaborator behaviour). Our controller is based on a unified task formalism, and in contrast with classical hybrid visicn–force–position control, it enables smooth transitions and weighted combinations of the sensor tasks. The approach is validated in a mock-up industrial scenario, where pose, vision (from both traditional camera and Kinect), and force tasks must be realized either exclusively or simultaneously, for human–robot collaboration.

Reinforcement learning when a human is the one providing the rewards to the algorithm

W. Bradley Knox, Peter Stone, Framing reinforcement learning from human reward: Reward positivity, temporal discounting, episodicity, and performance, Artificial Intelligence, Volume 225, August 2015, Pages 24-50, ISSN 0004-3702, DOI: 10.1016/j.artint.2015.03.009.

Several studies have demonstrated that reward from a human trainer can be a powerful feedback signal for control-learning algorithms. However, the space of algorithms for learning from such human reward has hitherto not been explored systematically. Using model-based reinforcement learning from human reward, this article investigates the problem of learning from human reward through six experiments, focusing on the relationships between reward positivity, which is how generally positive a trainer’s reward values are; temporal discounting, the extent to which future reward is discounted in value; episodicity, whether task learning occurs in discrete learning episodes instead of one continuing session; and task performance, the agent’s performance on the task the trainer intends to teach. This investigation is motivated by the observation that an agent can pursue different learning objectives, leading to different resulting behaviors. We search for learning objectives that lead the agent to behave as the trainer intends.
We identify and empirically support a “positive circuits” problem with low discounting (i.e., high discount factors) for episodic, goal-based tasks that arises from an observed bias among humans towards giving positive reward, resulting in an endorsement of myopic learning for such domains. We then show that converting simple episodic tasks to be non-episodic (i.e., continuing) reduces and in some cases resolves issues present in episodic tasks with generally positive reward and—relatedly—enables highly successful learning with non-myopic valuation in multiple user studies. The primary learning algorithm introduced in this article, which we call “vi-tamer”, is the first algorithm to successfully learn non-myopically from reward generated by a human trainer; we also empirically show that such non-myopic valuation facilitates higher-level understanding of the task. Anticipating the complexity of real-world problems, we perform further studies—one with a failure state added—that compare (1) learning when states are updated asynchronously with local bias—i.e., states quickly reachable from the agent’s current state are updated more often than other states—to (2) learning with the fully synchronous sweeps across each state in the vi-tamer algorithm. With these locally biased updates, we find that the general positivity of human reward creates problems even for continuing tasks, revealing a distinct research challenge for future work.

Heuristic, real-time search with weighted heuristic function

Nicolás Rivera, Jorge A. Baier, Carlos Hernández, Incorporating weights into real-time heuristic search, Artificial Intelligence, Volume 225, August 2015, Pages 1-23, ISSN 0004-3702, DOI: 10.1016/j.artint.2015.03.008.

Multiplying the heuristic function by a weight greater than one is a well-known technique in heuristic search. When this technique is applied to A* with an admissible heuristic it yields substantial runtime savings, at the expense of sacrificing solution optimality. Its applicability to real-time heuristic search, a search approach that builds upon heuristic search, however, has only been explored by a few studies. In this article we present two new approaches to using weights in real-time heuristic search, applicable to a wide range of algorithms. The first one, weighted lookahead, is a variant of an existing approach by Shimbo and Ishida, and utilizes the weight while the algorithm performs lookahead search. The second one, weighted update, incorporates the weight to the edges of the search graph during the learning phase. We implemented both techniques within LSS-LRTA* and evaluated them in path-planning benchmarks. We show that weighted lookahead outperforms an existing approach by Shimbo and Ishida but that it does not improve over existing approaches that do not use weights. Weighted update, on the other hand, yields performance improvements of up to one order of magnitude both in solution cost and total search time. To illustrate further the generality of weighted update, we incorporate the technique in two other well-known real-time heuristic search algorithms: LRTA*-LS and daLSS-LRTA*, and we empirically show significant improvements for LRTA*-LS and modest but still important improvements for daLSS-LRTA*. We analyze the properties of weighted update in depth, showing, among other things, that it guarantees termination. Convergence behavior of LSS-LRTA*, modified to use weighted update, is also analyzed. In such a setting, we prove solutions are w-optimal, and provide additional bounds on solution quality that in practice are tighter than w-optimality.