Header logo is


2017


Thumb xl fig toyex lqr1kernel 1
On the Design of LQR Kernels for Efficient Controller Learning

Marco, A., Hennig, P., Schaal, S., Trimpe, S.

Proceedings of the 56th IEEE Annual Conference on Decision and Control (CDC), pages: 5193-5200, IEEE, IEEE Conference on Decision and Control, December 2017 (conference)

Abstract
Finding optimal feedback controllers for nonlinear dynamic systems from data is hard. Recently, Bayesian optimization (BO) has been proposed as a powerful framework for direct controller tuning from experimental trials. For selecting the next query point and finding the global optimum, BO relies on a probabilistic description of the latent objective function, typically a Gaussian process (GP). As is shown herein, GPs with a common kernel choice can, however, lead to poor learning outcomes on standard quadratic control problems. For a first-order system, we construct two kernels that specifically leverage the structure of the well-known Linear Quadratic Regulator (LQR), yet retain the flexibility of Bayesian nonparametric learning. Simulations of uncertain linear and nonlinear systems demonstrate that the LQR kernels yield superior learning performance.

am ics pn

arXiv PDF On the Design of LQR Kernels for Efficient Controller Learning - CDC presentation DOI Project Page [BibTex]

2017


arXiv PDF On the Design of LQR Kernels for Efficient Controller Learning - CDC presentation DOI Project Page [BibTex]


Thumb xl probls sketch n3 0 ei0
Probabilistic Line Searches for Stochastic Optimization

Mahsereci, M., Hennig, P.

Journal of Machine Learning Research, 18(119):1-59, November 2017 (article)

pn

link (url) Project Page [BibTex]

link (url) Project Page [BibTex]


Thumb xl teaser
Coupling Adaptive Batch Sizes with Learning Rates

Balles, L., Romero, J., Hennig, P.

In Proceedings Conference on Uncertainty in Artificial Intelligence (UAI) 2017, pages: 410-419, (Editors: Gal Elidan and Kristian Kersting), Association for Uncertainty in Artificial Intelligence (AUAI), Conference on Uncertainty in Artificial Intelligence (UAI), August 2017 (inproceedings)

Abstract
Mini-batch stochastic gradient descent and variants thereof have become standard for large-scale empirical risk minimization like the training of neural networks. These methods are usually used with a constant batch size chosen by simple empirical inspection. The batch size significantly influences the behavior of the stochastic optimization algorithm, though, since it determines the variance of the gradient estimates. This variance also changes over the optimization process; when using a constant batch size, stability and convergence is thus often enforced by means of a (manually tuned) decreasing learning rate schedule. We propose a practical method for dynamic batch size adaptation. It estimates the variance of the stochastic gradients and adapts the batch size to decrease the variance proportionally to the value of the objective function, removing the need for the aforementioned learning rate decrease. In contrast to recent related work, our algorithm couples the batch size to the learning rate, directly reflecting the known relationship between the two. On three image classification benchmarks, our batch size adaptation yields faster optimization convergence, while simultaneously simplifying learning rate tuning. A TensorFlow implementation is available.

ps pn

Code link (url) Project Page [BibTex]

Code link (url) Project Page [BibTex]


no image
Dynamic Time-of-Flight

Schober, M., Adam, A., Yair, O., Mazor, S., Nowozin, S.

Proceedings IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2017, pages: 170-179, IEEE, Piscataway, NJ, USA, IEEE Conference on Computer Vision and Pattern Recognition (CVPR), July 2017 (conference)

ei pn

DOI [BibTex]

DOI [BibTex]


Thumb xl this one
Virtual vs. Real: Trading Off Simulations and Physical Experiments in Reinforcement Learning with Bayesian Optimization

Marco, A., Berkenkamp, F., Hennig, P., Schoellig, A. P., Krause, A., Schaal, S., Trimpe, S.

In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA), pages: 1557-1563, IEEE, Piscataway, NJ, USA, IEEE International Conference on Robotics and Automation (ICRA), May 2017 (inproceedings)

am ics pn

PDF arXiv ICRA 2017 Spotlight presentation Virtual vs. Real - Video explanation DOI Project Page [BibTex]

PDF arXiv ICRA 2017 Spotlight presentation Virtual vs. Real - Video explanation DOI Project Page [BibTex]


Thumb xl screen shot 2017 06 14 at 2.38.22 pm
Scalable Pneumatic and Tendon Driven Robotic Joint Inspired by Jumping Spiders

Sproewitz, A., Göttler, C., Sinha, A., Caer, C., Öztekin, M. U., Petersen, K., Sitti, M.

In Proceedings 2017 IEEE International Conference on Robotics and Automation (ICRA), pages: 64-70, IEEE, Piscataway, NJ, USA, IEEE International Conference on Robotics and Automation (ICRA), May 2017 (inproceedings)

dlg

Video link (url) DOI Project Page [BibTex]

Video link (url) DOI Project Page [BibTex]


Thumb xl screen shot 2017 07 20 at 12.31.00 pm
Fast Bayesian Optimization of Machine Learning Hyperparameters on Large Datasets

Klein, A., Falkner, S., Bartels, S., Hennig, P., Hutter, F.

Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS 2017), 54, pages: 528-536, Proceedings of Machine Learning Research, (Editors: Sign, Aarti and Zhu, Jerry), PMLR, April 2017 (conference)

pn

pdf link (url) Project Page [BibTex]

pdf link (url) Project Page [BibTex]


Thumb xl screen shot 2017 06 14 at 2.58.42 pm
Spinal joint compliance and actuation in a simulated bounding quadruped robot

Pouya, S., Khodabakhsh, M., Sproewitz, A., Ijspeert, A.

{Autonomous Robots}, pages: 437–452, Kluwer Academic Publishers, Springer, Dordrecht, New York, NY, Febuary 2017 (article)

dlg

link (url) DOI Project Page [BibTex]

link (url) DOI Project Page [BibTex]


Thumb xl early stopping teaser
Early Stopping Without a Validation Set

Mahsereci, M., Balles, L., Lassner, C., Hennig, P.

arXiv preprint arXiv:1703.09580, 2017 (article)

Abstract
Early stopping is a widely used technique to prevent poor generalization performance when training an over-expressive model by means of gradient-based optimization. To find a good point to halt the optimizer, a common practice is to split the dataset into a training and a smaller validation set to obtain an ongoing estimate of the generalization performance. In this paper we propose a novel early stopping criterion which is based on fast-to-compute, local statistics of the computed gradients and entirely removes the need for a held-out validation set. Our experiments show that this is a viable approach in the setting of least-squares and logistic regression as well as neural networks.

ps pn

link (url) Project Page Project Page [BibTex]


no image
Krylov Subspace Recycling for Fast Iterative Least-Squares in Machine Learning

Roos, F. D., Hennig, P.

arXiv preprint arXiv:1706.00241, 2017 (article)

Abstract
Solving symmetric positive definite linear problems is a fundamental computational task in machine learning. The exact solution, famously, is cubicly expensive in the size of the matrix. To alleviate this problem, several linear-time approximations, such as spectral and inducing-point methods, have been suggested and are now in wide use. These are low-rank approximations that choose the low-rank space a priori and do not refine it over time. While this allows linear cost in the data-set size, it also causes a finite, uncorrected approximation error. Authors from numerical linear algebra have explored ways to iteratively refine such low-rank approximations, at a cost of a small number of matrix-vector multiplications. This idea is particularly interesting in the many situations in machine learning where one has to solve a sequence of related symmetric positive definite linear problems. From the machine learning perspective, such deflation methods can be interpreted as transfer learning of a low-rank approximation across a time-series of numerical tasks. We study the use of such methods for our field. Our empirical results show that, on regression and classification problems of intermediate size, this approach can interpolate between low computational cost and numerical precision.

pn

link (url) Project Page [BibTex]


no image
Convergence Analysis of Deterministic Kernel-Based Quadrature Rules in Misspecified Settings

Kanagawa, M., Sriperumbudur, B. K., Fukumizu, K.

Arxiv e-prints, arXiv:1709.00147v1 [math.NA], 2017 (article)

Abstract
This paper presents convergence analysis of kernel-based quadrature rules in misspecified settings, focusing on deterministic quadrature in Sobolev spaces. In particular, we deal with misspecified settings where a test integrand is less smooth than a Sobolev RKHS based on which a quadrature rule is constructed. We provide convergence guarantees based on two different assumptions on a quadrature rule: one on quadrature weights, and the other on design points. More precisely, we show that convergence rates can be derived (i) if the sum of absolute weights remains constant (or does not increase quickly), or (ii) if the minimum distance between distance design points does not decrease very quickly. As a consequence of the latter result, we derive a rate of convergence for Bayesian quadrature in misspecified settings. We reveal a condition on design points to make Bayesian quadrature robust to misspecification, and show that, under this condition, it may adaptively achieve the optimal rate of convergence in the Sobolev space of a lesser order (i.e., of the unknown smoothness of a test integrand), under a slightly stronger regularity condition on the integrand.

pn

arXiv [BibTex]

arXiv [BibTex]


no image
Fast Bayesian hyperparameter optimization on large datasets

Klein, A., Falkner, S., Bartels, S., Hennig, P., Hutter, F.

Electronic Journal of Statistics, 11, 2017 (article)

pn

[BibTex]

[BibTex]


Thumb xl screen shot 2018 02 08 at 12.58.55 pm
Linking Mechanics and Learning

Heim, S., Grimminger, F., Özge, D., Spröwitz, A.

In Proceedings of Dynamic Walking 2017, 2017 (inproceedings)

dlg

[BibTex]

[BibTex]


no image
New Directions for Learning with Kernels and Gaussian Processes (Dagstuhl Seminar 16481)

Gretton, A., Hennig, P., Rasmussen, C., Schölkopf, B.

Dagstuhl Reports, 6(11):142-167, 2017 (book)

ei pn

DOI [BibTex]

DOI [BibTex]


Thumb xl screen shot 2018 02 08 at 12.58.55 pm
Is Growing Good for Learning?

Heim, S., Spröwitz, A.

Proceedings of the 8th International Symposium on Adaptive Motion of Animals and Machines AMAM2017, 2017 (conference)

dlg

[BibTex]

[BibTex]


no image
Efficiency of analytical and sampling-based uncertainty propagation in intensity-modulated proton therapy

Wahl, N., Hennig, P., Wieser, H. P., Bangert, M.

Physics in Medicine & Biology, 62(14):5790-5807, 2017 (article)

Abstract
The sensitivity of intensity-modulated proton therapy (IMPT) treatment plans to uncertainties can be quantified and mitigated with robust/min-max and stochastic/probabilistic treatment analysis and optimization techniques. Those methods usually rely on sparse random, importance, or worst-case sampling. Inevitably, this imposes a trade-off between computational speed and accuracy of the uncertainty propagation. Here, we investigate analytical probabilistic modeling (APM) as an alternative for uncertainty propagation and minimization in IMPT that does not rely on scenario sampling. APM propagates probability distributions over range and setup uncertainties via a Gaussian pencil-beam approximation into moments of the probability distributions over the resulting dose in closed form. It supports arbitrary correlation models and allows for efficient incorporation of fractionation effects regarding random and systematic errors. We evaluate the trade-off between run-time and accuracy of APM uncertainty computations on three patient datasets. Results are compared against reference computations facilitating importance and random sampling. Two approximation techniques to accelerate uncertainty propagation and minimization based on probabilistic treatment plan optimization are presented. Runtimes are measured on CPU and GPU platforms, dosimetric accuracy is quantified in comparison to a sampling-based benchmark (5000 random samples). APM accurately propagates range and setup uncertainties into dose uncertainties at competitive run-times (GPU ##IMG## [http://ej.iop.org/images/0031-9155/62/14/5790/pmbaa6ec5ieqn001.gif] {$\leqslant {5}$} min). The resulting standard deviation (expectation value) of dose show average global ##IMG## [http://ej.iop.org/images/0031-9155/62/14/5790/pmbaa6ec5ieqn002.gif] {$\gamma_{{3}\% / {3}~{\rm mm}}$} pass rates between 94.2% and 99.9% (98.4% and 100.0%). All investigated importance sampling strategies provided less accuracy at higher run-times considering only a single fraction. Considering fractionation, APM uncertainty propagation and treatment plan optimization was proven to be possible at constant time complexity, while run-times of sampling-based computations are linear in the number of fractions. Using sum sampling within APM, uncertainty propagation can only be accelerated at the cost of reduced accuracy in variance calculations. For probabilistic plan optimization, we were able to approximate the necessary pre-computations within seconds, yielding treatment plans of similar quality as gained from exact uncertainty propagation. APM is suited to enhance the trade-off between speed and accuracy in uncertainty propagation and probabilistic treatment plan optimization, especially in the context of fractionation. This brings fully-fledged APM computations within reach of clinical application.

pn

link (url) [BibTex]

link (url) [BibTex]


no image
Analytical probabilistic modeling of RBE-weighted dose for ion therapy

Wieser, H., Hennig, P., Wahl, N., Bangert, M.

Physics in Medicine and Biology (PMB), 62(23):8959-8982, 2017 (article)

pn

link (url) [BibTex]

link (url) [BibTex]


Thumb xl screen shot 2018 02 08 at 1.12.35 pm
Evaluation of the passive dynamics of compliant legs with inertia

Györfi, B.

University of Applied Science Pforzheim, Germany, 2017 (mastersthesis)

dlg

[BibTex]

[BibTex]

2010


no image
Using an Infinite Von Mises-Fisher Mixture Model to Cluster Treatment Beam Directions in External Radiation Therapy

Bangert, M., Hennig, P., Oelfke, U.

In pages: 746-751 , (Editors: Draghici, S. , T.M. Khoshgoftaar, V. Palade, W. Pedrycz, M.A. Wani, X. Zhu), IEEE, Piscataway, NJ, USA, Ninth International Conference on Machine Learning and Applications (ICMLA), December 2010 (inproceedings)

Abstract
We present a method for fully automated selection of treatment beam ensembles for external radiation therapy. We reformulate the beam angle selection problem as a clustering problem of locally ideal beam orientations distributed on the unit sphere. For this purpose we construct an infinite mixture of von Mises-Fisher distributions, which is suited in general for density estimation from data on the D-dimensional sphere. Using a nonparametric Dirichlet process prior, our model infers probability distributions over both the number of clusters and their parameter values. We describe an efficient Markov chain Monte Carlo inference algorithm for posterior inference from experimental data in this model. The performance of the suggested beam angle selection framework is illustrated for one intra-cranial, pancreas, and prostate case each. The infinite von Mises-Fisher mixture model (iMFMM) creates between 18 and 32 clusters, depending on the patient anatomy. This suggests to use the iMFMM directly for beam ensemble selection in robotic radio surgery, or to generate low-dimensional input for both subsequent optimization of trajectories for arc therapy and beam ensemble selection for conventional radiation therapy.

ei pn

Web DOI [BibTex]

2010


Web DOI [BibTex]


no image
Approximate Inference in Graphical Models

Hennig, P.

University of Cambridge, November 2010 (phdthesis)

ei pn

Web [BibTex]

Web [BibTex]


no image
Coherent Inference on Optimal Play in Game Trees

Hennig, P., Stern, D., Graepel, T.

In JMLR Workshop and Conference Proceedings Volume 9: AISTATS 2010, pages: 326-333, (Editors: Teh, Y.W. , M. Titterington ), JMLR, Cambridge, MA, USA, Thirteenth International Conference on Artificial Intelligence and Statistics, May 2010 (inproceedings)

Abstract
Round-based games are an instance of discrete planning problems. Some of the best contemporary game tree search algorithms use random roll-outs as data. Relying on a good policy, they learn on-policy values by propagating information upwards in the tree, but not between sibling nodes. Here, we present a generative model and a corresponding approximate message passing scheme for inference on the optimal, off-policy value of nodes in smooth AND/OR trees, given random roll-outs. The crucial insight is that the distribution of values in game trees is not completely arbitrary. We define a generative model of the on-policy values using a latent score for each state, representing the value under the random roll-out policy. Inference on the values under the optimal policy separates into an inductive, pre-data step and a deductive, post-data part. Both can be solved approximately with Expectation Propagation, allowing off-policy value inference for any node in the (exponentially big) tree in linear time.

ei pn

PDF Web [BibTex]

PDF Web [BibTex]


Thumb xl screen shot 2018 02 03 at 4.33.15 pm
Graph signature for self-reconfiguration planning of modules with symmetry

Asadpour, M., Ashtiani, M. H. Z., Spröwitz, A., Ijspeert, A. J.

In Proceedings of the 2009 IEEE/RSJ International Conference on Intelligent Robots and Systems, pages: 5295-5300, IEEE, St. Louis, MO, 2010 (inproceedings)

Abstract
In our previous works we had developed a framework for self-reconfiguration planning based on graph signature and graph edit-distance. The graph signature is a fast isomorphism test between different configurations and the graph edit-distance is a similarity metric. But the algorithm is not suitable for modules with symmetry. In this paper we improve the algorithm in order to deal with symmetric modules. Also, we present a new heuristic function to guide the search strategy by penalizing the solutions with more number of actions. The simulation results show the new algorithm not only deals with symmetric modules successfully but also finds better solutions in a shorter time.

dlg

DOI [BibTex]

DOI [BibTex]


Thumb xl screen shot 2018 02 03 at 11.59.15 am
Roombots - Towards decentralized reconfiguration with self-reconfiguring modular robotic metamodules

Spröwitz, A., Laprade, P., Bonardi, S., Mayer, M., Moeckel, R., Mudry, P., Ijspeert, A. J.

In Proceedings of the 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages: 1126-1132, IEEE, Taipeh, 2010 (inproceedings)

Abstract
This paper presents our work towards a decentralized reconfiguration strategy for self-reconfiguring modular robots, assembling furniture-like structures from Roombots (RB) metamodules. We explore how reconfiguration by loco- motion from a configuration A to a configuration B can be controlled in a distributed fashion. This is done using Roombots metamodules—two Roombots modules connected serially—that use broadcast signals, lookup tables of their movement space, assumptions about their neighborhood, and connections to a structured surface to collectively build desired structures without the need of a centralized planner.

dlg

DOI [BibTex]

DOI [BibTex]


Thumb xl screen shot 2018 02 03 at 7.21.10 pm
Roombots: Reconfigurable Robots for Adaptive Furniture

Spröwitz, A., Pouya, S., Bonardi, S., van den Kieboom, J., Möckel, R., Billard, A., Dillenbourg, P., Ijspeert, A.

Computational Intelligence Magazine, IEEE, 5(3):20-32, 2010 (article)

Abstract
Imagine a world in which our furniture moves around like legged robots, interacts with us, and changes shape and function during the day according to our needs. This is the long term vision we have in the Roombots project. To work towards this dream, we are developing modular robotic modules that have rotational degrees of freedom for locomotion as well as active connection mechanisms for runtime reconfiguration. A piece of furniture, e.g. a stool, will thus be composed of several modules that activate their rotational joints together to implement locomotor gaits, and will be able to change shape, e.g. transforming into a chair, by sequences of attachments and detachments of modules. In this article, we firstly present the project and the hardware we are currently developing. We explore how reconfiguration from a configuration A to a configuration B can be controlled in a distributed fashion. This is done using metamodules-two Roombots modules connected serially-that use broadcast signals and connections to a structured ground to collectively build desired structures without the need of a centralized planner. We then present how locomotion controllers can be implemented in a distributed system of coupled oscillators-one per degree of freedom-similarly to the concept of central pattern generators (CPGs) found in the spinal cord of vertebrate animals. The CPGs are based on coupled phase oscillators to ensure synchronized behavior and have different output filters to allow switching between oscillations and rotations. A stochastic optimization algorithm is used to explore optimal CPG configurations for different simulated Roombots structures.

dlg

DOI [BibTex]

DOI [BibTex]


Thumb xl screen shot 2018 02 03 at 4.24.19 pm
Distributed Online Learning of Central Pattern Generators in Modular Robots

Christensen, D. J., Spröwitz, A., Ijspeert, A. J.

In From Animals to Animats 11, 6226, pages: 402-412, Lecture Notes in Computer Science, Springer, Berlin, 2010, author: Doncieux, Stéphan (incollection)

Abstract
In this paper we study distributed online learning of locomotion gaits for modular robots. The learning is based on a stochastic ap- proximation method, SPSA, which optimizes the parameters of coupled oscillators used to generate periodic actuation patterns. The strategy is implemented in a distributed fashion, based on a globally shared reward signal, but otherwise utilizing local communication only. In a physics-based simulation of modular Roombots robots we experiment with online learn- ing of gaits and study the effects of: module failures, different robot morphologies, and rough terrains. The experiments demonstrate fast online learning, typically 5-30 min. for convergence to high performing gaits (≈ 30 cm/sec), despite high numbers of open parameters (45-54). We conclude that the proposed approach is efficient, effective and a promising candidate for online learning on many other robotic platforms.

dlg

DOI [BibTex]

DOI [BibTex]


Thumb xl screen shot 2018 02 03 at 4.38.20 pm
Automatic Gait Generation in Modular Robots: to Oscillate or to Rotate? that is the question

Pouya, S., van den Kieboom, J., Spröwitz, A., Ijspeert, A. J.

In Proceedings of the 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages: 514-520, IEEE, Taipei, 2010 (inproceedings)

Abstract
Modular robots offer the possibility to design robots with a high diversity of shapes and functionalities. This nice feature also brings an important challenge: namely how to design efficient locomotion gaits for arbitrary robot structures with many degrees of freedom. In this paper, we present a framework that allows one to explore and identify highly different gaits for a given arbitrary- shaped modular robot. We use simulated robots made of several Roombots modules that have three rotational joints each. These modules have the interesting feature that they can produce both oscillatory movements (i.e. periodic movements around a rest position) and rotational movements (i.e. with continuously increasing angle), leading to very rich locomotion patterns. Here we ask ourselves which types of movements —purely oscillatory, purely rotational, or a combination of both— lead to the fastest gaits. To address this question we designed a control architecture based on a distributed system of coupled phase oscillators that can produce synchronized rotations and oscillations in many degrees of freedom. We also designed a specific optimization algorithm that can automatically design hybrid controllers, i.e. controllers that use oscillations in some joints and rotations in others, for fast gaits. The proposed framework is verified by multiple simulations for several robot morphologies. The results show that (i) the question whether it is better to oscillate or to rotate depends on the morphology of the robot, and that in general it is best to do both, (ii) the optimization framework can successfully generate hybrid controllers that outperform purely oscillatory and purely rotational ones, and (iii) the resulting gaits are fast, innovative, and would have been hard to design by hand.

dlg

DOI [BibTex]

DOI [BibTex]


Thumb xl screen shot 2018 02 03 at 4.30.26 pm
Roombots: Design and Implementation of a Modular Robot for Reconfiguration and Locomotion

Spröwitz, A.

EPFL, Lausanne, Lausanne, 2010 (phdthesis)

dlg

DOI [BibTex]