Refine
Year of publication
Document Type
- Preprint (759)
- Article (402)
- Working Paper (119)
- Doctoral Thesis (93)
- Diploma Thesis (47)
- Conference Proceeding (41)
- Book (37)
- Bachelor Thesis (36)
- diplomthesis (28)
- Report (25)
Has Fulltext
- yes (1619)
Is part of the Bibliography
- no (1619)
Keywords
Institute
- Informatik (1619) (remove)
The Internet as the biggest human library ever assembled keeps on growing. Although all kinds of information carriers (e.g. audio/video/hybrid file formats) are available, text based documents dominate. It is estimated that about 80% of all information worldwide stored electronically exists in (or can be converted into) text form. More and more, all kinds of documents are generated by means of a text processing system and are therefore available electronically. Nowadays, many printed journals are also published online and may even discontinue to appear in print form tomorrow. This development has many convincing advantages: the documents are both available faster (cf. prepress services) and cheaper, they can be searched more easily, the physical storage only needs a fraction of the space previously necessary and the medium will not age. For most people, fast and easy access is the most interesting feature of the new age; computer-aided search for specific documents or Web pages becomes the basic tool for information-oriented work. But this tool has problems. The current keyword based search machines available on the Internet are not really appropriate for such a task; either there are (way) too many documents matching the specified keywords are presented or none at all. The problem lies in the fact that it is often very difficult to choose appropriate terms describing the desired topic in the first place. This contribution discusses the current state-of-the-art techniques in content-based searching (along with common visualization/browsing approaches) and proposes a particular adaptive solution for intuitive Internet document navigation, which not only enables the user to provide full texts instead of manually selected keywords (if available), but also allows him/her to explore the whole database.
Efficient algorithms for object recognition are crucial for the newly robotics and computer vision applications that demand real-time and on-line methods. Some examples are autonomous systems, navigating robots, autonomous driving. In this work, we focus on efficient semantic segmentation, which is the problem of labeling each pixel of an image with a semantic class.
Our aim is to speed-up all of the parts of the semantic segmentation pipeline. We also aim at delivering a labeling solution on a time budget, that can be decided on-the-fly. For this purpose, we analyze all the components of the semantic segmentation pipeline, and identify the computational bottleneck of each of them. The different components of the pipeline are over-segmenting the image with local regions, extracting features and classify the local regions, and the final inference of the image labeling with semantic classes. We focus on each of these steps.
First, we introduce a new superpixel algorithm to over-segment the image. Our superpixel method runs in real-time and can deliver a solution at any time budget. Then, for feature extraction, we focus on the framework that computes descriptors and encodes them, followed by a pooling step. We see that the encoding step is the bottleneck, for computational efficiency and performance. We present a novel assignment-based encoding formulation, that allows for the design of a new, very efficient, encoding. Finally, the image labeling output is obtained modeling the dependencies with a Conditional Random Field (CRF). In semantic image segmentation, the computational cost of instantiating the potentials is much higher than MAP inference. We introduce Active MAP inference to on-the-fly select a subset of potentials to be instantiated in the energy function, leaving the rest as unknown, and to estimate the MAP labeling from such incomplete energy function.
We perform experiments on all proposed methods for the different parts of the semantic segmentation pipeline. We show that our superpixel extraction achieves higher accuracy than state-of-the-art on standard superpixel benchmark, while it runs in real-time. We test our feature encoding on standard image classification and segmentation benchmarks, and we show that our method achieves competitive results with the state-of-the-art, and requires less time and memory. Finally, results for semantic segmentation benchmark show that Active MAP inference achieves similar levels of accuracy but with major efficiency gains.
Active efficient coding explains the development of binocular vision and its failure in amblyopia
(2020)
The development of vision during the first months of life is an active process that comprises the learning of appropriate neural representations and the learning of accurate eye movements. While it has long been suspected that the two learning processes are coupled, there is still no widely accepted theoretical framework describing this joint development. Here, we propose a computational model of the development of active binocular vision to fill this gap. The model is based on a formulation of the active efficient coding theory, which proposes that eye movements as well as stimulus encoding are jointly adapted to maximize the overall coding efficiency. Under healthy conditions, the model self-calibrates to perform accurate vergence and accommodation eye movements. It exploits disparity cues to deduce the direction of defocus, which leads to coordinated vergence and accommodation responses. In a simulated anisometropic case, where the refraction power of the two eyes differs, an amblyopia-like state develops in which the foveal region of one eye is suppressed due to inputs from the other eye. After correcting for refractive errors, the model can only reach healthy performance levels if receptive fields are still plastic, in line with findings on a critical period for binocular vision development. Overall, our model offers a unifying conceptual framework for understanding the development of binocular vision.
The interaction between Λ baryons and kaons/antikaons is a crucial ingredient for the strangeness S=0 and S=−2 sector of the meson--baryon interaction at low energies. In particular, the ΛK¯¯¯¯ might help in understanding the origin of states such as the Ξ(1620), whose nature and properties are still under debate. Experimental data on Λ−K and Λ−K¯¯¯¯ systems are scarce, leading to large uncertainties and tension between the available theoretical predictions constrained by such data. In this Letter we present the measurements of Λ−K+⊕Λ¯¯¯¯−K− and Λ−K−⊕Λ¯¯¯¯−K+ correlations obtained in the high-multiplicity triggered data sample in pp collisions at s√=13 TeV recorded by ALICE at the LHC. The correlation function for both pairs is modeled using the Lednicky−Lyuboshits analytical formula and the corresponding scattering parameters are extracted. The Λ−K−⊕Λ¯¯¯¯−K+ correlations show the presence of several structures at relative momenta k∗ above 200 MeV/c, compatible with the Ω baryon, the Ξ(1690), and Ξ(1820) resonances decaying into Λ−K− pairs. The low k∗ region in the Λ−K−⊕Λ¯¯¯¯−K+ also exhibits the presence of the Ξ(1620) state, expected to strongly couple to the measured pair. The presented data allow to access the ΛK+ and ΛK− strong interaction with an unprecedented precision and deliver the first experimental observation of the Ξ(1620) decaying into ΛK−.
The interaction between Λ baryons and kaons/antikaons is a crucial ingredient for the strangeness S=0 and S=−2 sector of the meson−baryon interaction at low energies. In particular, the ΛK¯¯¯¯ might help in understanding the origin of states such as the Ξ(1620), whose nature and properties are still under debate. Experimental data on Λ−K and Λ−K¯¯¯¯ systems are scarce, leading to large uncertainties and tension between the available theoretical predictions constrained by such data. In this Letter we present the measurements of Λ−K+⊕Λ¯¯¯¯−K− and Λ−K−⊕Λ¯¯¯¯−K+ correlations obtained in the high-multiplicity triggered data sample in pp collisions at s√=13 TeV recorded by ALICE at the LHC. The correlation function for both pairs is modeled using the Lednicky−Lyuboshits analytical formula and the corresponding scattering parameters are extracted. The Λ−K−⊕Λ¯¯¯¯−K+ correlations show the presence of several structures at relative momenta k∗ above 200 MeV/c, compatible with the Ω baryon, the Ξ(1690), and Ξ(1820) resonances decaying into Λ−K− pairs. The low k∗ region in the Λ−K−⊕Λ¯¯¯¯−K+ also exhibits the presence of the Ξ(1620) state, expected to strongly couple to the measured pair. The presented data allow to access the ΛK+ and ΛK− strong interaction with an unprecedented precision and deliver the first experimental observation of the Ξ(1620) decaying into ΛK−.
The interaction between Λ baryons and kaons/antikaons is a crucial ingredient for the strangeness S=0 and S=−2 sector of the meson–baryon interaction at low energies. In particular, the ΛK‾ might help in understanding the origin of states such as the Ξ(1620), whose nature and properties are still under debate. Experimental data on Λ–K and Λ–K‾ systems are scarce, leading to large uncertainties and tension between the available theoretical predictions constrained by such data. In this Letter we present the measurements of Λ–K⊕+Λ‾–K− and Λ–K⊕−Λ‾–K+ correlations obtained in the high-multiplicity triggered data sample in pp collisions at s=13 TeV recorded by ALICE at the LHC. The correlation function for both pairs is modeled using the Lednický–Lyuboshits analytical formula and the corresponding scattering parameters are extracted. The Λ–K⊕−Λ‾–K+ correlations show the presence of several structures at relative momenta k⁎ above 200 MeV/c, compatible with the Ω baryon, the Ξ(1690), and Ξ(1820) resonances decaying into Λ–K− pairs. The low k⁎ region in the Λ–K⊕−Λ‾–K+ also exhibits the presence of the Ξ(1620) state, expected to strongly couple to the measured pair. The presented data allow to access the ΛK+ and ΛK− strong interaction with an unprecedented precision and deliver the first experimental observation of the Ξ(1620) decaying into ΛK−.
The interaction between Λ baryons and kaons/antikaons is a crucial ingredient for the strangeness S=0 and S=−2 sector of the meson--baryon interaction at low energies. In particular, the ΛK¯¯¯¯ might help in understanding the origin of states such as the Ξ(1620), whose nature and properties are still under debate. Experimental data on Λ−K and Λ−K¯¯¯¯ systems are scarce, leading to large uncertainties and tension between the available theoretical predictions constrained by such data. In this Letter we present the measurements of Λ−K+⊕Λ¯¯¯¯−K− and Λ−K−⊕Λ¯¯¯¯−K+ correlations obtained in the high-multiplicity triggered data sample in pp collisions at s√=13 TeV recorded by ALICE at the LHC. The correlation function for both pairs is modeled using the Lednicky−Lyuboshits analytical formula and the corresponding scattering parameters are extracted. The Λ−K−⊕Λ¯¯¯¯−K+ correlations show the presence of several structures at relative momenta k∗ above 200 MeV/c, compatible with the Ω baryon, the Ξ(1690), and Ξ(1820) resonances decaying into Λ−K− pairs. The low k∗ region in the Λ−K−⊕Λ¯¯¯¯−K+ also exhibits the presence of the Ξ(1620) state, expected to strongly couple to the measured pair. The presented data allow to access the ΛK+ and ΛK− strong interaction with an unprecedented precision and deliver the first experimental observation of the Ξ(1620) decaying into ΛK−.
Acceleration of Biomedical Image Processing and Reconstruction with FPGAs
Increasing chip sizes and better programming tools have made it possible to increase the boundaries of application acceleration with reconfigurable computer chips. In this thesis the potential of acceleration with Field Programmable Gate Arrays (FPGAs) is examined for applications that perform biomedical image processing and reconstruction. The dataflow paradigm was used to port the analysis of image data for localization microscopy and for 3D electron tomography from an imperative description towards the FPGA for the first time.
After the primitives of image processing on FPGAs are presented, a general workflow is given for analyzing imperative source code and converting it to a hardware pipeline where every node processes image data in parallel. The theoretical foundation is then used to accelerate both example applications. For localization microscopy, an acceleration of 185 compared to an Intel i5 450 CPU was achieved, and electron tomography could be sped up by a factor of 5 over an Nvidia Tesla C1060 graphics card while maintaining full accuracy in both cases.
We present the FPGA implementation of an algorithm [4] that computes implications between signal values in a boolean network. The research was performed as a masterrsquos thesis [5] at the University of Frankfurt. The recursive algorithm is rather complex for a hardware realization and therefore the FPGA implementation is an interesting example for the potential of reconfigurable computing beyond systolic algorithms. A circuit generator was written that transforms a boolean network into a network of small processing elements and a global control logic which together implement the algorithm. The resulting circuit performs the computation two orders of magnitudes faster than a software implementation run by a conventional workstation.
The early prediction of mortality is one of the unresolved tasks in intensive care medicine. This contribution models medical symptoms as observations cased by transitions between hidden markov states. Learning the underlying state transition probabilities results in a prediction probability success of about 91%. The results are discussed and put in relation to the model used. Finally, the rationales for using the model are reflected: Are there states in the septic shock data?
Current deep learning methods are regarded as favorable if they empirically perform well on dedicated test sets. This mentality is seamlessly reflected in the resurfacing area of continual learning, where consecutively arriving data is investigated. The core challenge is framed as protecting previously acquired representations from being catastrophically forgotten. However, comparison of individual methods is nevertheless performed in isolation from the real world by monitoring accumulated benchmark test set performance. The closed world assumption remains predominant, i.e. models are evaluated on data that is guaranteed to originate from the same distribution as used for training. This poses a massive challenge as neural networks are well known to provide overconfident false predictions on unknown and corrupted instances. In this work we critically survey the literature and argue that notable lessons from open set recognition, identifying unknown examples outside of the observed set, and the adjacent field of active learning, querying data to maximize the expected performance gain, are frequently overlooked in the deep learning era. Hence, we propose a consolidated view to bridge continual learning, active learning and open set recognition in deep neural networks. Finally, the established synergies are supported empirically, showing joint improvement in alleviating catastrophic forgetting, querying data, selecting task orders, while exhibiting robust open world application.
Natural Language Processing (NLP) for big data requires an efficient and sophisticated infrastructure to complete tasks both fast and correctly. Providing an intuitive and lightweight interaction with a framework that abstracts and simplifies complex tasks assists in reaching this goal. This bachelor thesis extends the NLP framework Docker Unified UIMA Interface (DUUI) by an API and a web-based graphical user interface to control and manage pipelines for automated analysis of large quantities of natural language. The extension aims to reduce the entry barrier into the field as well as to accelerate the creation and management of pipelines according to UIMA standards. Pipelines can be executed in the browser or using the web API directly and then monitored on a document level. The evaluation in usability and user experience indicates that the implementation benefits the framework by making its usage more user friendly, lightweight, and intuitive while also making the management of pipelines more efficient.
One of the most interesting domains of feedforward networks is the processing of sensor signals. There do exist some networks which extract most of the information by implementing the maximum entropy principle for Gaussian sources. This is done by transforming input patterns to the base of eigenvectors of the input autocorrelation matrix with the biggest eigenvalues. The basic building block of these networks is the linear neuron, learning with the Oja learning rule. Nevertheless, some researchers in pattern recognition theory claim that for pattern recognition and classification clustering transformations are needed which reduce the intra-class entropy. This leads to stable, reliable features and is implemented for Gaussian sources by a linear transformation using the eigenvectors with the smallest eigenvalues. In another paper (Brause 1992) it is shown that the basic building block for such a transformation can be implemented by a linear neuron using an Anti-Hebb rule and restricted weights. This paper shows the analog VLSI design for such a building block, using standard modules of multiplication and addition. The most tedious problem in this VLSI-application is the design of an analog vector normalization circuitry. It can be shown that the standard approaches of weight summation will not give the convergence to the eigenvectors for a proper feature transformation. To avoid this problem, our design differs significantly from the standard approaches by computing the real Euclidean norm. Keywords: minimum entropy, principal component analysis, VLSI, neural networks, surface approximation, cluster transformation, weight normalization circuit.
The well-known proof of termination of reduction in simply typed calculi is adapted to a monomorphically typed lambda-calculus with case and constructors and recursive data types. The proof differs at several places from the standard proof. Perhaps it is useful and can be extended also to more complex calculi.
We introduce a new method for representing and solving a general class of non-preemptive resource-constrained project scheduling problems. The new approach is to represent scheduling problems as de- scriptions (activity terms) in a language called RSV, which allows nested expressions using pll, seq, and xor. The activity-terms of RSV are similar to concepts in a description logic. The language RSV generalizes previous approaches to scheduling with variants insofar as it permits xor's not only of atomic activities but also of arbitrary activity terms. A specific semantics that assigns their set of active schedules to activity terms shows correctness of a calculus normalizing activity terms RSV similar to propositional DNF-computation. Based on RSV, this paper describes a diagram-based algorithm for the RSV-problem which uses a scan-line principle. The scan-line principle is used for determining and resolving the occurring resource conflicts and leads to a nonredundant generation of all active schedules and thus to a computation of the optimal schedule.
Non-coding variations located within regulatory elements may alter gene expression by modifying Transcription Factor (TF) binding sites and thereby lead to functional consequences like various traits or diseases. To understand these molecular mechanisms, different TF models are being used to assess the effect of DNA sequence variations, such as Single Nucleotide Polymorphisms (SNPs). However, few statistical approaches exist to compute statistical significance of results but they often are slow for large sets of SNPs, such as data obtained from a genome-wide association study (GWAS) or allele-specific analysis of chromatin data.
Results We investigate the distribution of maximal differential TF binding scores for general computational models that assess TF binding. We find that a modified Laplace distribution can adequately approximate the empirical distributions. A benchmark on in vitro and in vivo data sets showed that our new approach improves on an existing method in terms of performance and speed. In applications on large sets of eQTL and GWAS SNPs we could illustrate the usefulness of the novel statistic to highlight cell type specific regulators and TF target genes.
Conclusions Our approach allows the evaluation of DNA changes that induce differential TF binding in a fast and accurate manner, permitting computations on large mutation data sets. An implementation of the novel approach is freely available at https://github.com/SchulzLab/SNEEP.
We study the following problem: given x element Rn either find a short integer relation m element Zn, so that =0 holds for the inner product <.,.>, or prove that no short integer relation exists for x. Hastad, Just Lagarias and Schnorr (1989) give a polynomial time algorithm for the problem. We present a stable variation of the HJLS--algorithm that preserves lower bounds on lambda(x) for infinitesimal changes of x. Given x \in {\RR}^n and \alpha \in \NN this algorithm finds a nearby point x' and a short integer relation m for x'. The nearby point x' is 'good' in the sense that no very short relation exists for points \bar{x} within half the x'--distance from x. On the other hand if x'=x then m is, up to a factor 2^{n/2}, a shortest integer relation for \mbox{x.} Our algorithm uses, for arbitrary real input x, at most \mbox{O(n^4(n+\log \alpha))} many arithmetical operations on real numbers. If x is rational the algorithm operates on integers having at most \mbox{O(n^5+n^3 (\log \alpha)^2 + \log (\|q x\|^2))} many bits where q is the common denominator for x.
After a short introduction into traditional image transform coding, multirate systems and multiscale signal coding the paper focuses on the subject of image encoding by a neural network. Taking also noise into account a network model is proposed which not only learns the optimal localized basis functions for the transform but also learns to implement a whitening filter by multi-resolution encoding. A simulation showing the multi-resolution capabilitys concludes the contribution.
Poster presentation: Introduction Dopaminergic neurons in the midbrain show a variety of firing patterns, ranging from very regular firing pacemaker cells to bursty and irregular neurons. The effects of different experimental conditions (like pharmacological treatment or genetical manipulations) on these neuronal discharge patterns may be subtle. Applying a stochastic model is a quantitative approach to reveal these changes. ...
A partial rehabilitation of side-effecting I/O : non-determinism in non-strict functional languages
(1996)
We investigate the extension of non-strict functional languages like Haskell or Clean by a non-deterministic interaction with the external world. Using call-by-need and a natural semantics which describes the reduction of graphs, this can be done such that the Church-Rosser Theorems 1 and 2 hold. Our operational semantics is a base to recognise which particular equivalencies are preserved by program transformations. The amount of sequentialisation may be smaller than that enforced by other approaches and the programming style is closer to the common one of side-effecting programming. However, not all program transformations used by an optimising compiler for Haskell remain correct in all contexts. Our result can be interpreted as a possibility to extend current I/O-mechanism by non-deterministic deterministic memoryless function calls. For example, this permits a call to a random number generator. Adding memoryless function calls to monadic I/O is possible and has a potential to extend the Haskell I/O-system.