Image novelty detection is a repeating task in computer vision and describes the detection of anomalous images based on a training dataset consisting solely of normal reference data. It has been ...found that, in particular, neural networks are well-suited for the task. Our approach first transforms the training and test images into ensembles of patches, which enables the assessment of mean-shifts between normal data and outliers. As mean-shifts are only detectable when the outlier ensemble and inlier distribution are spatially separate from each other, a rich feature space, such as a pre-trained neural network, needs to be chosen to represent the extracted patches. For mean-shift estimation, the Hotelling T2 test is used. The size of the patches turned out to be a crucial hyperparameter that needs additional domain knowledge about the spatial size of the expected anomalies (local vs. global). This also affects model selection and the chosen feature space, as commonly used Convolutional Neural Networks or Vision Image Transformers have very different receptive field sizes. To showcase the state-of-the-art capabilities of our approach, we compare results with classical and deep learning methods on the popular dataset CIFAR-10, and demonstrate its real-world applicability in a large-scale industrial inspection scenario using the MVTec dataset. Because of the inexpensive design, our method can be implemented by a single additional 2D-convolution and pooling layer and allows particularly fast prediction times while being very data-efficient.
The strain in a fast spinning carbon fiber flywheel rotor is of great interest for condition monitoring, as well as for studying long-term aging effects in the carbon fiber matrix. Optoelectronic ...strain measurement is a contactless measurement principle where a special reflective pattern is applied to the rotor which is scanned by a stationary optical setup. It does not require any active electronic components on the rotor and is suited for operation in a vacuum. In this paper, the influences of the key parts comprising the optoelectronic strain measurement are analyzed. The influence of each part on the measurement result including the uncertainty is modeled. The total uncertainty, as well as each part's contribution is calculated. This provides a valuable assessment of requirements for component selection, as well as tolerances of mechanical parts and processes to reach a final target measurement uncertainty or to estimate the uncertainty of a given setup. We have shown that the edge quality of the special reflective pattern has the strongest influence, and how to improve it. Considering all influences, it is possible to measure strain with an uncertainty of less than 1% at a rotation speed of 500Hz.
Monitoring the strain in the rotating flywheel in a kinetic energy storage system is important for safe operation and for the investigation of long-term effects in composite materials like ...carbon-fiber-reinforced plastics. An optoelectronic strain-measurement system for contactless deformation and position monitoring of a flywheel was investigated. The system consists of multiple optical sensors measuring the local relative in-plane displacement of the flywheel rotor. A special reflective pattern, which is necessary to interact with the sensors, was applied to the surface of the rotor. Combining the measurements from multiple sensors makes it possible to distinguish between the deformation and in-plane displacement of the flywheel. The sensor system was evaluated using a low-speed steel rotor for single-sensor performance investigation as well as a scaled-down high-speed rotor made from PVC plastic. The PVC rotor exhibits more deformation due to centrifugal stresses than a steel or aluminum rotor of the same dimensions, which allows experimental measurements at a smaller flywheel scale as well as a lower rotation speed. Deformation measurements were compared to expected deformation from calculations. The influence of sensor distance was investigated. Deformation and position measurements as well as derived imbalance measurements were demonstrated.
We are interested in computing a mini-batch-capable end-to-end algorithm to identify statistically independent components (ICA) in large scale and high-dimensional datasets. Current algorithms ...typically rely on pre-whitened data and do not integrate the two procedures of whitening and ICA estimation. Our online approach estimates a whitening and a rotation matrix with stochastic gradient descent on centered or uncentered data. We show that this can be done efficiently by combining Batch Karhunen-Löwe-Transformation 1 with Lie group techniques. Our algorithm is recursion-free and can be organized as feed-forward neural network which makes the use of GPU acceleration straight-forward. Because of the very fast convergence of Batch KLT, the gradient descent in the Lie group of orthogonal matrices stabilizes quickly. The optimization is further enhanced by integrating ADAM 2, an improved stochastic gradient descent (SGD) technique from the field of deep learning. We test the scaling capabilities by computing the independent components of the well-known ImageNet challenge (144 GB). Due to its robustness with respect to batch and step size, our approach can be used as a drop-in replacement for standard ICA algorithms where memory is a limiting factor.
Facial mimicry is the automatic imitation of the facial affect expressions of others. It serves as an important component of interpersonal communication and affective co-experience. Facial mimicry ...has so far been measured by Electromyography (EMG), which requires a complex measuring apparatus. Recently, software for measuring facial expressions have become available, but it is still unclear how well it is suited for measuring facial mimicry. This study investigates the comparability of the automated facial coding software Affdex with EMG for measuring facial mimicry. For this purpose, facial mimicry was induced in 33 subjects by presenting naturalistic affect-expressive video sequences (anger, joy). The response of the subjects is measured simultaneously by facial EMG (corrugator supercilii muscle, zygomaticus major muscle) and by Affdex (action units lip corner puller and brow lowerer and affects joy and anger). Subsequently, the correlations between the measurement results of EMG and Affdex were calculated. After the presentation of the joy stimulus, there was an increase in zygomaticus muscle activity (EMG) about 400 ms after stimulus onset and an increase in joy and lip corner puller activity (Affdex) about 1200 ms after stimulus onset. The joy and the lip corner puller activity detected by Affdex correlate significantly with the EMG activity. After presentation of the anger stimulus, corrugator muscle activity (EMG) also increased approximately 400 ms after stimulus onset, whereas anger and brow lowerer activity (Affdex) showed no response. During the entire measurement interval, anger activity and brow lowerer activity (Affdex) did not correlate with corrugator muscle activity (EMG). Using Affdex, the facial mimicry response to a joy stimulus can be measured, but it is detected approximately 800 ms later compared to the EMG. Thus, electromyography remains the tool of choice for studying subtle mimic processes like facial mimicry.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
We formulate the notion of an isomorphism of GKM graphs. We then show that two GKM graphs have isomorphic graph equivariant cohomology algebras if and only if the graphs are isomorphic.
Echolocating bats use the echoes from their echolocation calls to perceive their surroundings. The ability to use these continuously emitted calls, whose main function is not communication, for ...recognition of individual conspecifics might facilitate many of the social behaviours observed in bats. Several studies of individual-specific information in echolocation calls found some evidence for its existence but did not quantify or explain it. We used a direct paradigm to show that greater mouse-eared bats (Myotis myotis) can easily discriminate between individuals based on their echolocation calls and that they can generalize their knowledge to discriminate new individuals that they were not trained to recognize. We conclude that, despite their high variability, broadband bat-echolocation calls contain individual-specific information that is sufficient for recognition. An analysis of the call spectra showed that formant-related features are suitable cues for individual recognition. As a model for the bat's decision strategy, we trained nonlinear statistical classifiers to reproduce the behaviour of the bats, namely to repeat correct and incorrect decisions of the bats. The comparison of the bats with the model strongly implies that the bats are using a prototype classification approach: they learn the average call characteristics of individuals and use them as a reference for classification.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Volterra and Wiener series are perhaps the best-understood nonlinear system representations in signal processing. Although both approaches have enjoyed a certain popularity in the past, their ...application has been limited to rather low-dimensional and weakly nonlinear systems due to the exponential growth of the number of terms that have to be estimated. We show that Volterra and Wiener series can be represented implicitly as elements of a reproducing kernel Hilbert space by using polynomial kernels. The estimation complexity of the implicit representation is linear in the input dimensionality and independent of the degree of nonlinearity. Experiments show performance advantages in terms of convergence, interpretability, and system sizes that can be handled.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
A critical step on the way to understanding a sensory system is the analysis of the input it receives. In this work we examine the statistics of natural complex echoes, focusing on vegetation echoes. ...Vegetation echoes constitute a major part of the sensory world of more than 800 species of echolocating bats and play an important role in several of their daily tasks. Our statistical analysis is based on a large collection of plant echoes acquired by a biomimetic sonar system. We explore the relation between the physical world (the structure of the plant) and the characteristics of its echo. Finally, we complete the story by analyzing the effect of the sensory processing of both the echolocation and the auditory systems on the echoes and interpret them in the light of information maximization. The echoes of all different plant species we examined share a surprisingly robust pattern that was also reproduced by a simple Poisson model of the spatial reflector arrangement. The fine differences observed between the echoes of different plant species can be explained by the spatial characteristics of the plants. The bat's emitted signal enhances the most informative spatial frequency range where the species-specific information is large. The auditory system filtering affects the echoes in a similar way, thus enhancing the most informative spatial frequency range even more. These findings suggest how the bat's sensory system could have evolved to deal with complex natural echoes.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
PDF
