Technical University of Berlin

About: Technical University of Berlin is a education organization based out in Berlin, Germany. It is known for research contribution in the topics: Laser & Catalysis. The organization has 27292 authors who have published 59342 publications receiving 1414623 citations. The organization is also known as: Technische Universität Berlin & TU Berlin.

A Benchmark for Interpretability Methods in Deep Neural Networks

Abstract: We propose an empirical measure of the approximate accuracy of feature importance estimates in deep neural networks. Our results across several large-scale image classification datasets show that many popular interpretability methods produce estimates of feature importance that are not better than a random designation of feature importance. Only certain ensemble based approaches---VarGrad and SmoothGrad-Squared---outperform such a random assignment of importance. The manner of ensembling remains critical, we show that some approaches do no better then the underlying method but carry a far higher computational burden.

Efficient Neural Audio Synthesis

Abstract: Sequential models achieve state-of-the-art results in audio, visual and textual domains with respect to both estimating the data distribution and generating high-quality samples. Efficient sampling for this class of models has however remained an elusive problem. With a focus on text-to-speech synthesis, we describe a set of general techniques for reducing sampling time while maintaining high output quality. We first describe a single-layer recurrent neural network, the WaveRNN, with a dual softmax layer that matches the quality of the state-of-the-art WaveNet model. The compact form of the network makes it possible to generate 24kHz 16-bit audio 4x faster than real time on a GPU. Second, we apply a weight pruning technique to reduce the number of weights in the WaveRNN. We find that, for a constant number of parameters, large sparse networks perform better than small dense networks and this relationship holds for sparsity levels beyond 96%. The small number of weights in a Sparse WaveRNN makes it possible to sample high-fidelity audio on a mobile CPU in real time. Finally, we propose a new generation scheme based on subscaling that folds a long sequence into a batch of shorter sequences and allows one to generate multiple samples at once. The Subscale WaveRNN produces 16 samples per step without loss of quality and offers an orthogonal method for increasing sampling efficiency.

A Human--Exoskeleton Interface Utilizing Electromyography

Abstract: This paper presents a human--machine interface to control exoskeletons that utilizes electrical signals from the muscles of the operator as the main means of information transportation. These signals are recorded with electrodes attached to the skin on top of selected muscles and reflect the activation of the observed muscle. They are evaluated by a sophisticated but simplified biomechanical model of the human body to derive the desired action of the operator. A support action is computed in accordance to the desired action and is executed by the exoskeleton. The biomechanical model fuses results from different biomechanical and biomedical research groups and performs a sensible simplification considering the intended application. Some of the model parameters reflect properties of the individual human operator and his or her current body state. A calibration algorithm for these parameters is presented that relies exclusively on sensors mounted on the exoskeleton. An exoskeleton for knee joint support was designed and constructed to verify the model and to investigate the interaction between operator and machine in experiments with force support during everyday movements.