• Publications
  • Influence
Local Gaussian Process Regression for Real Time Online Model Learning
TLDR
This work proposes a method to speed up standard Gaussian process regression with local GP models (LGP), which has higher accuracy than LWPR and close to the performance of standard GPR and ν-SVR.
Evaluation of Policy Gradient Methods and Variants on the Cart-Pole Benchmark
TLDR
This paper evaluates different versions from the three main kinds of model-free policy gradient methods, i.e., finite difference gradients, 'vanilla' policy gradients and natural policy gradient methods, using the cart pole regulator benchmark.
Closing the sensorimotor loop: haptic feedback facilitates decoding of motor imagery.
TLDR
Empirical evidence is presented that haptic feedback, provided by a seven degrees of freedom robotic arm, facilitates online decoding of arm movement intention, which supports the feasibility of future rehabilitative treatments based on the combination of robot-assisted physical therapy with BCIs.
Learning Robot Dynamics for Computed Torque Control Using Local Gaussian Processes Regression
TLDR
Comparisons with other nonparametric regressions show that LGP has higher accuracy than LWPR and close to the performance of standard GPR and v-SVR while being sufficiently fast for online learning.
Learning object-specific grasp affordance densities
TLDR
The result of learning grasp hypothesis densities from both imitation and visual cues are shown, and grasp empirical densities learned from physical experience by a robot are presented.
Model-Based Relative Entropy Stochastic Search
TLDR
This work introduces a new surrogate-based stochastic search approach that considerably outperforms the existing approaches and uses information theoretic constraints to bound the 'distance' between the new and old data distribution while maximizing the objective function.
Sparse online model learning for robot control with support vector regression
TLDR
This paper proposes an approximation of the support vector regression (SVR) by sparsification based on the linear independency of training data which is applicable in real-time online learning and exhibits competitive learning accuracy when compared with standard regression techniques.
Recurrent Spiking Networks Solve Planning Tasks
TLDR
The neural planning method with its local update rules provides the basis for future neuromorphic hardware implementations with promising potentials like large data processing abilities and early initiation of strategies to avoid dangerous situations in robot co-worker scenarios.
Sharing Knowledge in Multi-Task Deep Reinforcement Learning
TLDR
This work studies the benefit of sharing representations among tasks to enable the effective use of deep neural networks in Multi-Task Reinforcement Learning, and extends the well-known finite-time bounds of Approximate Value-Iteration to the multi-task setting.
Switching Linear Dynamics for Variational Bayes Filtering
TLDR
Leveraging Bayesian inference, Variational Autoencoders and Concrete relaxations, it is shown how to learn a richer and more meaningful state space, e.g. encoding joint constraints and collisions with walls in a maze, from partial and high-dimensional observations.
...
...