Klaus-Robert Müller

Bio: Klaus-Robert Müller is an academic researcher from Technical University of Berlin. The author has contributed to research in topics: Artificial neural network & Support vector machine. The author has an hindex of 129, co-authored 764 publications receiving 79391 citations. Previous affiliations of Klaus-Robert Müller include Korea University & University of Tokyo.

Interpretable human action recognition in compressed domain

Abstract: Compressed domain human action recognition algorithms are extremely efficient, because they only require a partial decoding of the video bit stream. However, the question what exactly makes these algorithms decide for a particular action is still a mystery. In this paper, we present a general method, Layer-wise Relevance Propagation (LRP), to understand and interpret action recognition algorithms and apply it to a state-of-the-art compressed domain method based on Fisher vector encoding and SVM classification. By using LRP, the classifiers decisions are propagated back every step in the action recognition pipeline until the input is reached. This methodology allows to identify where and when the important (from the classifier's perspective) action happens in the video. To our knowledge, this is the first work to interpret a compressed domain action recognition algorithm. We evaluate our method on the HMDB51 dataset and show that in many cases a few significant frames contribute most towards the prediction of the video to a particular class.

Brain-Switches for Asynchronous Brain−Computer Interfaces: A Systematic Review

Abstract: A brain–computer interface (BCI) has been extensively studied to develop a novel communication system for disabled people using their brain activities. An asynchronous BCI system is more realistic and practical than a synchronous BCI system, in that, BCI commands can be generated whenever the user wants. However, the relatively low performance of an asynchronous BCI system is problematic because redundant BCI commands are required to correct false-positive operations. To significantly reduce the number of false-positive operations of an asynchronous BCI system, a two-step approach has been proposed using a brain-switch that first determines whether the user wants to use an asynchronous BCI system before the operation of the asynchronous BCI system. This study presents a systematic review of the state-of-the-art brain-switch techniques and future research directions. To this end, we reviewed brain-switch research articles published from 2000 to 2019 in terms of their (a) neuroimaging modality, (b) paradigm, (c) operation algorithm, and (d) performance.

StructRank: a new approach for ligand-based virtual screening.

Abstract: Screening large libraries of chemical compounds against a biological target, typically a receptor or an enzyme, is a crucial step in the process of drug discovery. Virtual screening (VS) can be seen as a ranking problem which prefers as many actives as possible at the top of the ranking. As a standard, current Quantitative Structure−Activity Relationship (QSAR) models apply regression methods to predict the level of activity for each molecule and then sort them to establish the ranking. In this paper, we propose a top-k ranking algorithm (StructRank) based on Support Vector Machines to solve the early recognition problem directly. Empirically, we show that our ranking approach outperforms not only regression methods but another ranking approach recently proposed for QSAR ranking, RankSVM, in terms of actives found.

Approximate Tree Kernels

Abstract: Convolution kernels for trees provide simple means for learning with tree-structured data. The computation time of tree kernels is quadratic in the size of the trees, since all pairs of nodes need to be compared. Thus, large parse trees, obtained from HTML documents or structured network data, render convolution kernels inapplicable. In this article, we propose an effective approximation technique for parse tree kernels. The approximate tree kernels (ATKs) limit kernel computation to a sparse subset of relevant subtrees and discard redundant structures, such that training and testing of kernel-based learning methods are significantly accelerated. We devise linear programming approaches for identifying such subsets for supervised and unsupervised learning tasks, respectively. Empirically, the approximate tree kernels attain run-time improvements up to three orders of magnitude while preserving the predictive accuracy of regular tree kernels. For unsupervised tasks, the approximate tree kernels even lead to more accurate predictions by identifying relevant dimensions in feature space.

Common spatial pattern patches - An optimized filter ensemble for adaptive brain-computer interfaces

Abstract: Laplacian filters are commonly used in Brain Computer Interfacing (BCI). When only data from few channels are available, or when, like at the beginning of an experiment, no previous data from the same user is available complex features cannot be used. In this case band power features calculated from Laplacian filtered channels represents an easy, robust and general feature to control a BCI, since its calculation does not involve any class information. For the same reason, the performance obtained with Laplacian features is poor in comparison to subject-specific optimized spatial filters, such as Common Spatial Patterns (CSP) analysis, which, on the other hand, can be used just in a later phase of the experiment, since they require a considerable amount of training data in order to enroll a stable and good performance. This drawback is particularly evident in case of poor performing BCI users, whose data is highly non-stationary and contains little class relevant information. Therefore, Laplacian filtering is preferred to CSP, e.g., in the initial period of co-adaptive calibration, a novel BCI paradigm designed to alleviate the problem of BCI illiteracy. In fact, in the co-adaptive calibration design the experiment starts with a subject-independent classifier and simple features are needed in order to obtain a fast adaptation of the classifier to the newly acquired user's data. Here, the use of an ensemble of local CSP patches (CSPP) is proposed, which can be considered as a compromise between Laplacians and CSP: CSPP needs less data and channels than CSP, while being superior to Laplacian filtering. This property is shown to be particularly useful for the co-adaptive calibration design and is demonstrated on off-line data from a previous co-adaptive BCI study.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Deep Learning

Abstract: Deep learning is a form of machine learning that enables computers to learn from experience and understand the world in terms of a hierarchy of concepts. Because the computer gathers knowledge from experience, there is no need for a human computer operator to formally specify all the knowledge that the computer needs. The hierarchy of concepts allows the computer to learn complicated concepts by building them out of simpler ones; a graph of these hierarchies would be many layers deep. This book introduces a broad range of topics in deep learning. The text offers mathematical and conceptual background, covering relevant concepts in linear algebra, probability theory and information theory, numerical computation, and machine learning. It describes deep learning techniques used by practitioners in industry, including deep feedforward networks, regularization, optimization algorithms, convolutional networks, sequence modeling, and practical methodology; and it surveys such applications as natural language processing, speech recognition, computer vision, online recommendation systems, bioinformatics, and videogames. Finally, the book offers research perspectives, covering such theoretical topics as linear factor models, autoencoders, representation learning, structured probabilistic models, Monte Carlo methods, the partition function, approximate inference, and deep generative models. Deep Learning can be used by undergraduate or graduate students planning careers in either industry or research, and by software engineers who want to begin using deep learning in their products or platforms. A website offers supplementary material for both readers and instructors.

I and i

Abstract: There is, I think, something ethereal about i —the square root of minus one. I remember first hearing about it at school. It seemed an odd beast at that time—an intruder hovering on the edge of reality. Usually familiarity dulls this sense of the bizarre, but in the case of i it was the reverse: over the years the sense of its surreal nature intensified. It seemed that it was impossible to write mathematics that described the real world in …

Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift

Abstract: Training Deep Neural Networks is complicated by the fact that the distribution of each layer's inputs changes during training, as the parameters of the previous layers change. This slows down the training by requiring lower learning rates and careful parameter initialization, and makes it notoriously hard to train models with saturating nonlinearities. We refer to this phenomenon as internal covariate shift, and address the problem by normalizing layer inputs. Our method draws its strength from making normalization a part of the model architecture and performing the normalization for each training mini-batch. Batch Normalization allows us to use much higher learning rates and be less careful about initialization, and in some cases eliminates the need for Dropout. Applied to a state-of-the-art image classification model, Batch Normalization achieves the same accuracy with 14 times fewer training steps, and beats the original model by a significant margin. Using an ensemble of batch-normalized networks, we improve upon the best published result on ImageNet classification: reaching 4.82% top-5 test error, exceeding the accuracy of human raters.