arXiv subjects
Publications and source records attributed to Christian Schulze.
We study the mixing properties of a passive scalar advected by an incompressible flow. We consider a class of cellular flows (more general than the class in [Crippa-Schulze M3AS 2017]) and show that, under the constraint that the palenstrophy is bounded uniformly in time, the mixing scale of the passive scalar cannot decay exponentially.
In the recent years, there has been a tremendous increase in the amount of video content uploaded to social networking and video sharing websites like Facebook and Youtube. As of result of this, the risk of children getting exposed to adult and violent content on the web also increased. To address this issue, an approach to automatically detect violent content in videos is proposed in this work. Here, a novel attempt is made also to detect the category of violence present in a video. A system which can automatically detect violence from both Hollywood movies and videos from the web is extremely useful not only in parental control but also for applications related to movie ratings, video surveillance, genre classification and so on. Here, both audio and visual features are used to detect violence. MFCC features are used as audio cues. Blood, Motion, and SentiBank features are used as visual cues. Binary SVM classifiers are trained on each of these features to detect violence. Late fusion using a weighted sum of classification scores is performed to get final classification scores for each of the violence class target by the system. To determine optimal weights for each of the violence classes an approach based on grid search is employed. Publicly available datasets, mainly Violent Scene Detection (VSD), are used for classifier training, weight calculation, and testing. The performance of the system is evaluated on two classification tasks, Multi-Class classification, and Binary Classification. The results obtained for Binary Classification are better than the baseline results from MediaEval-2014.
The detection of fraud in accounting data is a long-standing challenge in financial statement audits. Nowadays, the majority of applied techniques refer to handcrafted rules derived from known fraud scenarios. While fairly successful, these rules exhibit the drawback that they often fail to generalize beyond known fraud scenarios and fraudsters gradually find ways to circumvent them. In contrast, more advanced approaches inspired by the recent success of deep learning often lack seamless interpretability of the detected results. To overcome this challenge, we propose the application of adversarial autoencoder networks. We demonstrate that such artificial neural networks are capable of learning a semantic meaningful representation of real-world journal entries. The learned representation provides a holistic view on a given set of journal entries and significantly improves the interpretability of detected accounting anomalies. We show that such a representation combined with the networks reconstruction error can be utilized as an unsupervised and highly adaptive anomaly assessment. Experiments on two datasets and initial feedback received by forensic accountants underpinned the effectiveness of the approach.
We study the mixing properties of a scalar $ρ$ advected by a certain incompressible velocity field $u$ on the two dimensional unit ball, which is a stationary radial solution of the Euler equation. The scalar $ρ$ solves the continuity equation with velocity field $u$ and we can measure the degree of mixedness of~$ρ$ with two different scales commonly used in this setting, namely the geometric and the functional mixing scale. We develop a physical space approach well adapted for the quantitative analysis of the decay in time of the geometric mixing scale, which turns out to be polynomial for a large class of initial data. This extends previous results for the functional mixing scale, based on the explicit expression for the solution in Fourier variable, results that are also partially recovered by our approach.
Recently, sound recognition has been used to identify sounds, such as car and river. However, sounds have nuances that may be better described by adjective-noun pairs such as slow car, and verb-noun pairs such as flying insects, which are under explored. Therefore, in this work we investigate the relation between audio content and both adjective-noun pairs and verb-noun pairs. Due to the lack of datasets with these kinds of annotations, we collected and processed the AudioPairBank corpus consisting of a combined total of 1,123 pairs and over 33,000 audio files. One contribution is the previously unavailable documentation of the challenges and implications of collecting audio recordings with these type of labels. A second contribution is to show the degree of correlation between the audio content and the labels through sound recognition experiments, which yielded results of 70% accuracy, hence also providing a performance benchmark. The results and study in this paper encourage further exploration of the nuances in audio and are meant to complement similar research performed on images and text in multimedia analysis.
We study the problem of optimal mixing of a passive scalar $ρ$ advected by an incompressible flow on the two dimensional unit square. The scalar $ρ$ solves the continuity equation with a divergence-free velocity field $u$ with uniform-in-time bounds on the homogeneous Sobolev semi-norm $\dot{W}^{s,p}$, where $s>1$ and $1< p \leq \infty$. We measure the degree of mixedness of the tracer $ρ$ via the two different notions of mixing scale commonly used in this setting, namely the functional and the geometric mixing scale. For velocity fields with the above constraint, it is known that the decay of both mixing scales cannot be faster than exponential. Numerical simulations suggest that this exponential lower bound is in fact sharp, but so far there is no explicit analytical example which matches this result. We analyze velocity fields of cellular type, which is a special localized structure often used in constructions of explicit analytical examples of mixing flows and can be viewed as a generalization of the self-similar construction by Alberti, Crippa and Mazzucato. We show that for any velocity field of cellular type both mixing scales cannot decay faster than polynomially.
Controlling the modal content coupled into an optical fiber can be desireable in many situations, e.g. for adjusting the sensitivity of the guided field distribution to external perturbations. For this purpose we used a monolithic setup of a phase plate at a fiber inpute facet to excite selectivly higher order modes, which theoretically can provide a mode purity of more than 99%. We investigated the capabilities of this approach by complete modal decomposition of the fiber output signals, considering the achievable mode purity with respect to several possible imperfections of the setup. The experiments are compared with detailed numerical simulations and show a high agreement. Additionally a comparison with a well known setup with free space phase plates was undertaken. This showed the monolithic setup to be energetically twice as efficient.
There is significant interest in tailoring wave packets that transversely accelerate during propagation. The first realisation was in the optical domain, the Airy beam. Valid in the paraxial approximation, such beams were shown to transversely accelerate while following a parabolic caustic, even if the centroid itself travelled along a rectilinear path. Later, non-paraxial optical fields in the form of Weber and Mathieu beams were demonstrated, as well as non-optical wave packets with electrons. Such fields have found a plethora of applications from particle manipulation, spatial-temporal beam control, plasma control to non-linear optics. Here we demonstrate, for the first time, the angular acceleration of light, achieved by non-canonical superpositions of Bessel beams carrying orbital angular momentum. We show that by adjustment of a single parameter the acceleration and deceleration may be tuned continuously. We observe an unexpected energy transfer mechanism between spatial regions of the field which we investigate experimentally and theoretically. Our findings offer a new class of optical field that will enable studies in matter waves and opto-fluidics.
We present a real-time method to determine the beam propagation ratio M2 of laser beams. The all-optical measurement of modal amplitudes yields M2 parameters conform to the ISO standard method. The experimental technique is simple and fast, which allows to investigate laser beams under conditions inaccessible to other methods.
Thousands of different forms (words) are associated with thousands of different meanings (concepts) in a language computer model. Reasonable agreement with reality is found for the number of languages in a family and the Hamming distances between languages.
Urban segregation of different communities, like blacks and whites in the USA, has been simulated by Ising-like models since Schelling 1971. This research was accompanied by a scientific segregation, with sociologists and physicists ignoring each other until 2000. We review recent progress and also present some new two-temperature multi-cultural simulations.
When a region is conquered by people speaking another language, we assume within the Schulze model that at each iteration each person with probability s shifts to the conquering language. The time needed for the conquering language to become dominating is about 2/s for directed Barabasi-Albert networks, but diverges on the square lattice for decreasing s at some critical value sc
The Schelling model of 1971 is a complicated version of a square-lattice Ising model at zero temperature, to explain urban segregation, based on the neighbour preferences of the residents, without external reasons. Various versions between Ising and Schelling models give about the same results. Inhomogeneous "temperatures" T do not change the results much, while a feedback between segregation and T leads to a self-organisation of an average T.
An earlier study (Nettle 1999b) concluded, based on computer simulations and some inferences from empirical data, that languages will change the more slowly the larger the population gets. We replicate this study using a more complete language model for simulations (the Schulze model combined with a Barabasi-Albert net- work) and a richer empirical dataset (the World Atlas of Language Structures edited by Haspelmath et al. 2005). Our simulations show either a weak or stronger dependence of language change on population sizes depending on the parameter settings, and empirical data, like some of the simulations, show a weak dependence.
We present a model for diffusion in a molecularly crowded environment. The model consists of random barriers in percolation network. Random walks in the presence of slowly moving barriers show normal diffusion for long times, but anomalous diffusion at intermediate times. The effective exponents for square distance versus time usually are below one at these intermediate times, but can be also larger than one for high barrier concentrations. Thus we observe sub- as well as super-diffusion in a crowded environment.
Using the Schulze model for Monte Carlo simulations of language competition, we include a barrier between the top half and the bottom half of the lattice. We check under which conditions two different languages evolve as dominating in the two halves.
The recent availability of larger typological databases such as Haspelmath et al. (2005) has brought the linguistics community closer to having a solid, empirical foundation for making actual claims about prehistoric migrations, deep genealogical relationship, and patterns of areal linguistic interaction. Nevertheless, still more data are needed, and more than anything a number of methodological problems need to be adressed. These problems, which are the focus of this paper. These questions will be addressed from two perspectives. The first perspective is an empirical one, where observations primarily derive from analyses of the data of Haspelmath et al. (2005). The second perspective is a computational one, where simulations are drawn upon to test the effects of different parameters on the development of structural linguistic diversity. The results suggest that there is indeed some hope that we may derive new empirical insights regarding linguistic prehistory by drawing upon typological data. We do not, however, make any specific empirical claims in this paper, but instead concentrate our efforts on methodology
Our earlier language model is modified to allow for the survival of a minority language without higher status, just because of the pride of its speakers in their linguistic identity. An appendix studies the roughness of the interface for linguistic regions when one language conquers the whole territory.