FG Maschinelles Lernen

19 Items

Recent Submissions
Towards exact molecular dynamics simulations with invariant machine-learned models

Chmiela, Stefan (2019)

Molecular dynamics (MD) simulations constitute the cornerstone of contemporary atomistic modeling in chemistry, biology, and materials science. However, one of the widely recognized and increasingly pressing issues in MD simulations is the lack of accuracy of underlying classical interatomic potentials, which hinders truly predictive modeling of dynamics and function of (bio)molecular systems. ...

Features and machine learning systems for structured and sequential data

Schwenk, Guido (2019)

Modern web and communication technology relies heavily on sequential and structured data for its process execution and communication protocols. Due to its complex properties, a manual analysis and detection of problems on this data is too time-consuming and expensive, and hence not feasible. As a consequence, features and automatic learning systems on this type of data are highly sought after. ...

One-class classification in the presence of point, collective, and contextual anomalies

Görnitz, Nico (2019)

Anomaly detection has a prominent position in the processing pipeline of any real-world data-driven application. Its central goal is to detect and separate valid data points from malicious-anomalous-ones such that the cleaned data set can be processed further. In many applications, anomalies are even the prime objects of interest and need to be exposed early in order to avoid loss, e.g. in cred...

Opening the machine learning black box with Layer-wise Relevance Propagation

Lapuschkin, Sebastian (2019)

Machine learning techniques such as (Deep) Neural Networks are successfully solving a plethora of tasks, e.g. in image recognition and text analysis, and provide novel predictive models for complex physical, biological and chemical systems. However, due to the nested complex and non-linear structure of many machine learning models, this comes with the disadvantage of them acting as a black box,...

Intrusion Detection in Unlabeled Data with Quarter-sphere Support Vector Machines

Laskov, Pavel ; Schäfer, Christin ; Kotenko, Igor ; Müller, Klaus-Robert (2004)

The anomaly detection methods are receiving growing attention in the intrusion detection community. The two main reasons for this are their ability to handle large volumes of unlabeled data and to detect previously unknown attacks. In this contribution we investigate the application of a modern machine learning technique – one-class Support Vector Machines (SVM) – for anomaly detection in unlab...

Multimodal instrumentation and methods for neurotechnology out of the lab

Lühmann, Alexander von (2018)

In neuroscience and related fields, progress in instrumentation, computational power, and signal processing methods continuously provide novel and increasingly powerful tools toward the investigation of brain activity in real-time and everyday environments. Research into real-life and application-oriented, non-invasive neurotechnology bears a number of multidisciplinary challenges which need to...

Functional regression of densities with application to the simulation of molecular dynamics

Brockherde, Felix (2018)

Applications of machine learning have shown promising results modeling the non-interacting kinetic energy functional in 1-D. This holds the promise of enabling orbital-free density functional theory calculations, by-passing the computationally expensive Kohn-Sham equations. This would yield substantial savings in computer-time so that larger systems or longer time scales can be simulated. ...

Large-scale approximate EM-style learning and inference in generative graphical models for sparse coding

Shelton, Jacquelyn Ann (2018)

We propose a nonparametric procedure to achieve fast inference in generative graphical models when the number of latent states is very large. The approach is based on iterative latent variable preselection, where we alternate between learning a `selection function' to reveal the relevant latent variables, and using this to obtain a compact approximation of the posterior distribution for EM; thi...

Learning representations of atomistic systems with deep neural networks

Schütt, Kristof (2018)

Learning Representations of Atomistic Systems with Deep Neural Networks Deep Learning has been shown to learn efficient representations for structured data such as image, text or audio. However, with the rise of applying machine learning to quantum chemistry, research has been largely focused on the development of hand-crafted descriptors of atomistic systems. In this thesis, we propose novel n...

Explaining nonlinear classification decisions with deep Taylor decomposition

Montavon, Grégoire ; Lapuschkin, Sebastian ; Binder, Alexander ; Samek, Wojciech ; Müller, Klaus-Robert (2017-05)

Nonlinear methods such as Deep Neural Networks (DNNs) are the gold standard for various challenging machine learning problems such as image recognition. Although these methods perform impressively well, they have a significant disadvantage, the lack of transparency, limiting the interpretability of the solution and thus the scope of application in practice. Especially DNNs act as black boxes du...

Machine learning of accurate energy-conserving molecular force fields

Chmiela, Stefan ; Tkatchenko, Alexandre ; Sauceda, Huziel E. ; Poltavsky, Igor ; Schütt, Kristof T. ; Müller, Klaus-Robert (2017)

Using conservation of energy - a fundamental property of closed classical and quantum mechanical systems - we develop an efficient gradient-domain machine learning (GDML) approach to construct accurate molecular force fields using a restricted number of samples from ab initio molecular dynamics (AIMD) trajectories. The GDML implementation is able to reproduce global potential energy surfaces of...

Real-time robustness evaluation of regression based myoelectric control against arm position change and donning/doffing

Hwang, Han-Jeong ; Hahne, Janne Mathias ; Müller, Klaus-Robert (2017)

There are some practical factors, such as arm position change and donning/doffing, which prevent robust myoelectric control. The objective of this study is to precisely characterize the impacts of the two representative factors on myoelectric controllability in practical control situations, thereby providing useful references that can be potentially used to find better solutions for clinically ...

Learning from label proportions in brain-computer interfaces

Hübner, David ; Verhoeven, Thibault ; Schmid, Konstantin ; Müller, Klaus-Robert ; Tangermann, Michael ; Kindermans, Pieter-Jan (2017)

Objective Using traditional approaches, a brain-computer interface (BCI) requires the collection of calibration data for new subjects prior to online use. Calibration time can be reduced or eliminated e.g., by subject-to-subject transfer of a pre-trained classifier or unsupervised adaptive classification methods which learn from scratch and adapt over time. While such heuristics work well in p...

Universal exact algorithm for globally augmented MAP inference in structured prediction

Bauer, Alexander (2017)

The ultimate goal of discriminative learning is to train a prediction system by optimizing a desired measure of performance. Unlike in the standard learning scenario with univariate real-valued outputs, in structured prediction we aim at predicting a structured label corresponding to complex objects such as sequences, alignments, sets, or graphs. Here, structural support vector machine (SSVM) e...

Zero training for BCI – Reality for BCI systems based on event-related potentials

Tangermann, Michael ; Kindermans, Pieter-Jan ; Schreuder, Martijn ; Schrauwen, Benjamin ; Müller, Klaus-Robert (2013)

This contribution reviews how usability in Brain- Computer Interfaces (BCI) can be enhanced. As an example, an unsupervised signal processing approach is presented, which tackles usability by an algorithmic improvement from the field of machine learning. The approach completely omits the necessity of a calibration recording for BCIs based on event-related potential (ERP) paradigms. The positive...

The hybrid brain-computer interface: a bridge to assistive technology?

Müller-Putz, Gernot R. ; Schreuder, Martijn ; Tangermann, Michael ; Leeb, R. ; Millán del, R. J. (2013)

Brain-Computer Interfaces (BCIs) can be extended by other input signals to form a so-called hybrid BCI (hBCI). Such an hBCI allows the processing of several input signals with at least one brain signal for control purposes, i.e. communication and environmental control. This work shows the principle, technology and application of hBCIs and discusses future objectives.

Identifying causal networks of neuronal sources from EEG/MEG data with the phase slope index: a simulation study

Ewald, Arne ; Avarvand, Forooz Shahbazi ; Nolte, Guido (2013)

The investigation of functional neuronal synchronization has recently become a growing field of research. With high temporal resolution, electroencephalography and magnetoencephalography are well-suited measurement techniques to identify networks of interacting sources underlying the recorded data. The analysis of the data in terms of effective connectivity, nevertheless, contains intrinsic iss...

A critical assessment of the importance of seedling age in the system of rice intensification (sri) in Eastern India

Deb, Debal ; Lässig, Jörg ; Kloft, Marius (2012)

A survey of the system of rice intensification (SRI)-related literature indicates that different authors have drawn conflicting inferences about rice yield performances under the SRI, chiefly because the SRI methodology has been variously advocated, interpreted and implemented in the field using different rice varieties, seedling ages at transplantation, cultivation seasons and nutrient managem...

Transfer learning of gaits on a quadrupedal robot

Degrave, Jonas ; Burm, Michael ; Kindermans, Pieter-Jan ; Dambre, Joni ; Wyffels, Francis (2015)

Learning new gaits for compliant robots is a challenging multi-dimensional optimization task. Furthermore, to ensure optimal performance, the optimization process must be repeated for every variation in the environment, for example for every change in inclination of the terrain. This is unfortunately not possible using current approaches, since the time required for the optimization is simply t...