-
Bed-Attached Vibration Sensor System: A Machine Learning Approach for Fall Detection in Nursing Homes
Authors:
Thomas Bartz-Beielstein,
Axel Wellendorf,
Noah Pütz,
Jens Brandt,
Alexander Hinterleitner,
Richard Schulz,
Richard Scholz,
Olaf Mersmann,
Robin Knabe
Abstract:
The increasing shortage of nursing staff and the acute risk of falls in nursing homes pose significant challenges for the healthcare system. This study presents the development of an automated fall detection system integrated into care beds, aimed at enhancing patient safety without compromising privacy through wearables or video monitoring. Mechanical vibrations transmitted through the bed frame…
▽ More
The increasing shortage of nursing staff and the acute risk of falls in nursing homes pose significant challenges for the healthcare system. This study presents the development of an automated fall detection system integrated into care beds, aimed at enhancing patient safety without compromising privacy through wearables or video monitoring. Mechanical vibrations transmitted through the bed frame are processed using a short-time Fourier transform, enabling robust classification of distinct human fall patterns with a convolutional neural network. Challenges pertaining to the quantity and diversity of the data are addressed, proposing the generation of additional data with a specific emphasis on enhancing variation. While the model shows promising results in distinguishing fall events from noise using lab data, further testing in real-world environments is recommended for validation and improvement. Despite limited available data, the proposed system shows the potential for an accurate and rapid response to falls, mitigating health implications, and addressing the needs of an aging population. This case study was performed as part of the ZIM Project. Further research on sensors enhanced by artificial intelligence will be continued in the ShapeFuture Project.
△ Less
Submitted 6 December, 2024;
originally announced December 2024.
-
Enhancing Feature Selection and Interpretability in AI Regression Tasks Through Feature Attribution
Authors:
Alexander Hinterleitner,
Thomas Bartz-Beielstein,
Richard Schulz,
Sebastian Spengler,
Thomas Winter,
Christoph Leitenmeier
Abstract:
Research in Explainable Artificial Intelligence (XAI) is increasing, aiming to make deep learning models more transparent. Most XAI methods focus on justifying the decisions made by Artificial Intelligence (AI) systems in security-relevant applications. However, relatively little attention has been given to using these methods to improve the performance and robustness of deep learning algorithms.…
▽ More
Research in Explainable Artificial Intelligence (XAI) is increasing, aiming to make deep learning models more transparent. Most XAI methods focus on justifying the decisions made by Artificial Intelligence (AI) systems in security-relevant applications. However, relatively little attention has been given to using these methods to improve the performance and robustness of deep learning algorithms. Additionally, much of the existing XAI work primarily addresses classification problems. In this study, we investigate the potential of feature attribution methods to filter out uninformative features in input data for regression problems, thereby improving the accuracy and stability of predictions. We introduce a feature selection pipeline that combines Integrated Gradients with k-means clustering to select an optimal set of variables from the initial data space. To validate the effectiveness of this approach, we apply it to a real-world industrial problem - blade vibration analysis in the development process of turbo machinery.
△ Less
Submitted 25 September, 2024;
originally announced September 2024.
-
DualAD: Disentangling the Dynamic and Static World for End-to-End Driving
Authors:
Simon Doll,
Niklas Hanselmann,
Lukas Schneider,
Richard Schulz,
Marius Cordts,
Markus Enzweiler,
Hendrik P. A. Lensch
Abstract:
State-of-the-art approaches for autonomous driving integrate multiple sub-tasks of the overall driving task into a single pipeline that can be trained in an end-to-end fashion by passing latent representations between the different modules. In contrast to previous approaches that rely on a unified grid to represent the belief state of the scene, we propose dedicated representations to disentangle…
▽ More
State-of-the-art approaches for autonomous driving integrate multiple sub-tasks of the overall driving task into a single pipeline that can be trained in an end-to-end fashion by passing latent representations between the different modules. In contrast to previous approaches that rely on a unified grid to represent the belief state of the scene, we propose dedicated representations to disentangle dynamic agents and static scene elements. This allows us to explicitly compensate for the effect of both ego and object motion between consecutive time steps and to flexibly propagate the belief state through time. Furthermore, dynamic objects can not only attend to the input camera images, but also directly benefit from the inferred static scene structure via a novel dynamic-static cross-attention. Extensive experiments on the challenging nuScenes benchmark demonstrate the benefits of the proposed dual-stream design, especially for modelling highly dynamic agents in the scene, and highlight the improved temporal consistency of our approach. Our method titled DualAD not only outperforms independently trained single-task networks, but also improves over previous state-of-the-art end-to-end models by a large margin on all tasks along the functional chain of driving.
△ Less
Submitted 10 June, 2024;
originally announced June 2024.
-
S.T.A.R.-Track: Latent Motion Models for End-to-End 3D Object Tracking with Adaptive Spatio-Temporal Appearance Representations
Authors:
Simon Doll,
Niklas Hanselmann,
Lukas Schneider,
Richard Schulz,
Markus Enzweiler,
Hendrik P. A. Lensch
Abstract:
Following the tracking-by-attention paradigm, this paper introduces an object-centric, transformer-based framework for tracking in 3D. Traditional model-based tracking approaches incorporate the geometric effect of object- and ego motion between frames with a geometric motion model. Inspired by this, we propose S.T.A.R.-Track, which uses a novel latent motion model (LMM) to additionally adjust obj…
▽ More
Following the tracking-by-attention paradigm, this paper introduces an object-centric, transformer-based framework for tracking in 3D. Traditional model-based tracking approaches incorporate the geometric effect of object- and ego motion between frames with a geometric motion model. Inspired by this, we propose S.T.A.R.-Track, which uses a novel latent motion model (LMM) to additionally adjust object queries to account for changes in viewing direction and lighting conditions directly in the latent space, while still modeling the geometric motion explicitly. Combined with a novel learnable track embedding that aids in modeling the existence probability of tracks, this results in a generic tracking framework that can be integrated with any query-based detector. Extensive experiments on the nuScenes benchmark demonstrate the benefits of our approach, showing state-of-the-art performance for DETR3D-based trackers while drastically reducing the number of identity switches of tracks at the same time.
△ Less
Submitted 13 October, 2024; v1 submitted 30 June, 2023;
originally announced June 2023.
-
What is understandable in Bayesian network explanations?
Authors:
Raphaela Butz,
Renée Schulz,
Arjen Hommersom,
Marko van Eekelen
Abstract:
Explaining predictions from Bayesian networks, for example to physicians, is non-trivial. Various explanation methods for Bayesian network inference have appeared in literature, focusing on different aspects of the underlying reasoning. While there has been a lot of technical research, there is very little known about how well humans actually understand these explanations. In this paper, we presen…
▽ More
Explaining predictions from Bayesian networks, for example to physicians, is non-trivial. Various explanation methods for Bayesian network inference have appeared in literature, focusing on different aspects of the underlying reasoning. While there has been a lot of technical research, there is very little known about how well humans actually understand these explanations. In this paper, we present ongoing research in which four different explanation approaches were compared through a survey by asking a group of human participants to interpret the explanations.
△ Less
Submitted 4 October, 2021;
originally announced October 2021.
-
On Hard and Soft Decision Decoding of BCH Codes
Authors:
Martin Bossert,
Rebekka Schulz,
Sebastian Bitzer
Abstract:
The binary primitive BCH codes are cyclic and are constructed by choosing a subset of the cyclotomic cosets. Which subset is chosen determines the dimension, the minimum distance and the weight distribution of the BCH code. We construct possible BCH codes and determine their coderate, true minimum distance and the non-equivalent codes. A particular choice of cyclotomic cosets gives BCH codes which…
▽ More
The binary primitive BCH codes are cyclic and are constructed by choosing a subset of the cyclotomic cosets. Which subset is chosen determines the dimension, the minimum distance and the weight distribution of the BCH code. We construct possible BCH codes and determine their coderate, true minimum distance and the non-equivalent codes. A particular choice of cyclotomic cosets gives BCH codes which are, extended by one bit, equivalent to Reed-Muller codes, which is a known result from the sixties. We show that BCH codes have possibly better parameters than Reed-Muller codes, which are related in recent publications to polar codes. We study the decoding performance of these different BCH codes using information set decoding based on minimal weight codewords of the dual code. We show that information set decoding is possible even in case of a channel without reliability information since the decoding algorithm inherently calculates reliability information. Different BCH codes of the same rate are compared and different decoding performances and complexity are observed. Some examples of hard decision decoding of BCH codes have the same decoding performance as maximum likelihood decoding. All presented decoding methods can possibly be extended to include reliability information of a Gaussian channel for soft decision decoding. We show simulation results for soft decision list information set decoding and compare the performance to other methods.
△ Less
Submitted 20 June, 2022; v1 submitted 15 July, 2021;
originally announced July 2021.
-
Semantic Security for Indoor THz-Wireless Communication
Authors:
Rebekka Schulz,
Onur Günlü,
Robert Elschner,
Rafael F. Schaefer,
Carsten Schmidt-Langhorst,
Colja Schubert,
Robert F. H. Fischer
Abstract:
Physical-layer security (PLS) for industrial indoor terahertz (THz) wireless communication applications is considered. We use a similar model as being employed for additive white Gaussian noise (AWGN) wireless communication channels. A cell communication and a directed communication scenario are analyzed to illustrate the achievable semantic security guarantees for a wiretap channel with finite-bl…
▽ More
Physical-layer security (PLS) for industrial indoor terahertz (THz) wireless communication applications is considered. We use a similar model as being employed for additive white Gaussian noise (AWGN) wireless communication channels. A cell communication and a directed communication scenario are analyzed to illustrate the achievable semantic security guarantees for a wiretap channel with finite-blocklength THz-wireless communication links. We show that weakly directed transmitter (Alice) antennas, which allow cell-type communication with multiple legitimate receivers (Bobs) without adaptation of the alignment, result in large insecure regions. In the directed communication scenario, the resulting insecure regions are shown to cover a large volume of the indoor environment only if the distance between Alice and Bob is large. Thus, our results for the two selected scenarios reveal that there is a stringent trade-off between the targeted semantic security level and the number of reliably and securely accessible legitimate receivers. Furthermore, the effects of secrecy code parameters and antenna properties on the achievable semantic security levels are illustrated to show directions for possible improvements to guarantee practically-acceptable security levels with PLS methods for industrial indoor THz-wireless communication applications.
△ Less
Submitted 19 July, 2021; v1 submitted 20 May, 2021;
originally announced May 2021.
-
Place Categorization and Semantic Mapping on a Mobile Robot
Authors:
Niko Sünderhauf,
Feras Dayoub,
Sean McMahon,
Ben Talbot,
Ruth Schulz,
Peter Corke,
Gordon Wyeth,
Ben Upcroft,
Michael Milford
Abstract:
In this paper we focus on the challenging problem of place categorization and semantic mapping on a robot without environment-specific training. Motivated by their ongoing success in various visual recognition tasks, we build our system upon a state-of-the-art convolutional network. We overcome its closed-set limitations by complementing the network with a series of one-vs-all classifiers that can…
▽ More
In this paper we focus on the challenging problem of place categorization and semantic mapping on a robot without environment-specific training. Motivated by their ongoing success in various visual recognition tasks, we build our system upon a state-of-the-art convolutional network. We overcome its closed-set limitations by complementing the network with a series of one-vs-all classifiers that can learn to recognize new semantic classes online. Prior domain knowledge is incorporated by embedding the classification system into a Bayesian filter framework that also ensures temporal coherence. We evaluate the classification accuracy of the system on a robot that maps a variety of places on our campus in real-time. We show how semantic information can boost robotic object detection performance and how the semantic map can be used to modulate the robot's behaviour during navigation tasks. The system is made available to the community as a ROS module.
△ Less
Submitted 9 July, 2015;
originally announced July 2015.
-
Modelacion y Visualizacion Tridimensional Interactiva de Variables Electricas en Celdas de Electro-Obtencion con Electrodos Bipolares
Authors:
César Mena Labraña,
Ricardo Sánchez Schulz,
Lautaro Salazar Silva
Abstract:
The use of floating bipolar electrodes in electrowinning cells of copper constitutes a nonconventional technology that promises economic and operational impacts. This paper presents a computational tool for the simulation and analysis of such electrochemical cells. A new model is developed for floating electrodes and a method of finite difference is used to obtain the threedimensional distributi…
▽ More
The use of floating bipolar electrodes in electrowinning cells of copper constitutes a nonconventional technology that promises economic and operational impacts. This paper presents a computational tool for the simulation and analysis of such electrochemical cells. A new model is developed for floating electrodes and a method of finite difference is used to obtain the threedimensional distribution of the potential and the field of current density inside the cell. The analysis of the results is based on a technique for the interactive visualization of three-dimensional vectorial fields as lines of flow.
△ Less
Submitted 22 January, 2010; v1 submitted 22 January, 2010;
originally announced January 2010.
