arXiv subjects

Matthias Schubert

Publications and source records attributed to Matthias Schubert.

33 records · Page 2Linked to original sources

arXiv2022

V-Coder: Adaptive AutoEncoder for Semantic Disclosure in Knowledge Graphs

Semantic Web or Knowledge Graphs (KG) emerged to one of the most important information source for intelligent systems requiring access to structured knowledge. One of the major challenges is the extraction and processing of unambiguous information from textual data. Following the human perception, overlapping semantic linkages between two named entities become clear due to our common-sense about the context a relationship lives in which is not the case when we look at it from an automatically driven process of a machine. In this work, we are interested in the problem of Relational Resolution within the scope of KGs, i.e, we are investigating the inherent semantic of relationships between entities within a network. We propose a new adaptive AutoEncoder, called V-Coder, to identify relations inherently connecting entities from different domains. Those relations can be considered as being ambiguous and are candidates for disentanglement. Likewise to the Adaptive Learning Theory (ART), our model learns new patterns from the KG by increasing units in a competitive layer without discarding the previous observed patterns whilst learning the quality of each relation separately. The evaluation on real-world datasets of Freebase, Yago and NELL shows that the V-Coder is not only able to recover links from corrupted input data, but also shows that the semantic disclosure of relations in a KG show the tendency to improve link prediction. A semantic evaluation wraps the evaluation up.

cs.AI↗

arXiv2022

APPTeK: Agent-Based Predicate Prediction in Temporal Knowledge Graphs

In temporal Knowledge Graphs (tKGs), the temporal dimension is attached to facts in a knowledge base resulting in quadruples between entities such as (Nintendo, released, Super Mario, Sep-13-1985), where the predicate holds within a time interval or at a timestamp. We propose a reinforcement learning agent gathering temporal relevant information about the query entities' neighborhoods, simultaneously. We refer to the encodings of the explored graph structures as fingerprints which are used as input to a Q-network. Our agent decides sequentially which relation type needs to be explored next to expand the local subgraphs of the query entities. Our evaluation shows that the proposed method yields competitive results compared to state-of-the-art embedding algorithms for tKGs, and we additionally gain information about the relevant structures between subjects and objects.

cs.AI↗

arXiv2022

Box Supervised Video Segmentation Proposal Network

Video Object Segmentation (VOS) has been targeted by various fully-supervised and self-supervised approaches. While fully-supervised methods demonstrate excellent results, self-supervised ones, which do not use pixel-level ground truth, attract much attention. However, self-supervised approaches pose a significant performance gap. Box-level annotations provide a balanced compromise between labeling effort and result quality for image segmentation but have not been exploited for the video domain. In this work, we propose a box-supervised video object segmentation proposal network, which takes advantage of intrinsic video properties. Our method incorporates object motion in the following way: first, motion is computed using a bidirectional temporal difference and a novel bounding box-guided motion compensation. Second, we introduce a novel motion-aware affinity loss that encourages the network to predict positive pixel pairs if they share similar motion and color. The proposed method outperforms the state-of-the-art self-supervised benchmark by 16.4% and 6.9% $\mathcal{J}$ &$\mathcal{F}$ score and the majority of fully supervised methods on the DAVIS and Youtube-VOS dataset without imposing network architectural specifications. We provide extensive tests and ablations on the datasets, demonstrating the robustness of our method.

cs.CV↗

arXiv2021

Conifer Seedling Detection in UAV-Imagery with RGB-Depth Information

Monitoring of reforestation is currently being considerably streamlined through the use of drones and image recognition algorithms, which have already proven to be effective on colour imagery. In addition to colour imagery, elevation data is often also available. The primary aim of this work was to improve the performance of the faster-RCNN object detection algorithm by integrating this height information, which showed itself to notably improve performance. Interestingly, the structure of the network played a key role, with direct addition of the height information as a fourth image channel showing no improvement, while integration after the backbone network and before the region proposal network led to marked improvements. This effect persisted with very long training regimes. Increasing the resolution of this height information also showed little effect.

cs.CV↗

arXiv2021

SEA: Graph Shell Attention in Graph Neural Networks

A common issue in Graph Neural Networks (GNNs) is known as over-smoothing. By increasing the number of iterations within the message-passing of GNNs, the nodes' representations of the input graph align with each other and become indiscernible. Recently, it has been shown that increasing a model's complexity by integrating an attention mechanism yields more expressive architectures. This is majorly contributed to steering the nodes' representations only towards nodes that are more informative than others. Transformer models in combination with GNNs result in architectures including Graph Transformer Layers (GTL), where layers are entirely based on the attention operation. However, the calculation of a node's representation is still restricted to the computational working flow of a GNN. In our work, we relax the GNN architecture by means of implementing a routing heuristic. Specifically, the nodes' representations are routed to dedicated experts. Each expert calculates the representations according to their respective GNN workflow. The definitions of distinguishable GNNs result from k-localized views starting from the central node. We call this procedure Graph Shell Attention (SEA), where experts process different subgraphs in a transformer-motivated fashion. Intuitively, by increasing the number of experts, the models gain in expressiveness such that a node's representation is solely based on nodes that are located within the receptive field of an expert. We evaluate our architecture on various benchmark datasets showing competitive results compared to state-of-the-art models.

cs.LG↗

arXiv2020

Learning Self-Expression Metrics for Scalable and Inductive Subspace Clustering

Subspace clustering has established itself as a state-of-the-art approach to clustering high-dimensional data. In particular, methods relying on the self-expressiveness property have recently proved especially successful. However, they suffer from two major shortcomings: First, a quadratic-size coefficient matrix is learned directly, preventing these methods from scaling beyond small datasets. Secondly, the trained models are transductive and thus cannot be used to cluster out-of-sample data unseen during training. Instead of learning self-expression coefficients directly, we propose a novel metric learning approach to learn instead a subspace affinity function using a siamese neural network architecture. Consequently, our model benefits from a constant number of parameters and a constant-size memory footprint, allowing it to scale to considerably larger datasets. In addition, we can formally show that out model is still able to exactly recover subspace clusters given an independence assumption. The siamese architecture in combination with a novel geometric classifier further makes our model inductive, allowing it to cluster out-of-sample data. Additionally, non-linear clusters can be detected by simply adding an auto-encoder module to the architecture. The whole model can then be trained end-to-end in a self-supervised manner. This work in progress reports promising preliminary results on the MNIST dataset. In the spirit of reproducible research, me make all code publicly available. In future work we plan to investigate several extensions of our model and to expand experimental evaluation.

cs.LG↗

arXiv2020

Unsupervised Anomaly Detection for X-Ray Images

Obtaining labels for medical (image) data requires scarce and expensive experts. Moreover, due to ambiguous symptoms, single images rarely suffice to correctly diagnose a medical condition. Instead, it often requires to take additional background information such as the patient's medical history or test results into account. Hence, instead of focusing on uninterpretable black-box systems delivering an uncertain final diagnosis in an end-to-end-fashion, we investigate how unsupervised methods trained on images without anomalies can be used to assist doctors in evaluating X-ray images of hands. Our method increases the efficiency of making a diagnosis and reduces the risk of missing important regions. Therefore, we adopt state-of-the-art approaches for unsupervised learning to detect anomalies and show how the outputs of these methods can be explained. To reduce the effect of noise, which often can be mistaken for an anomaly, we introduce a powerful preprocessing pipeline. We provide an extensive evaluation of different approaches and demonstrate empirically that even without labels it is possible to achieve satisfying results on a real-world dataset of X-ray images of hands. We also evaluate the importance of preprocessing and one of our main findings is that without it, most of our approaches perform not better than random. To foster reproducibility and accelerate research we make our code publicly available at https://github.com/Valentyn1997/xray

eess.IV↗

arXiv2020

Small-Object Detection in Remote Sensing Images with End-to-End Edge-Enhanced GAN and Object Detector Network

The detection performance of small objects in remote sensing images is not satisfactory compared to large objects, especially in low-resolution and noisy images. A generative adversarial network (GAN)-based model called enhanced super-resolution GAN (ESRGAN) shows remarkable image enhancement performance, but reconstructed images miss high-frequency edge information. Therefore, object detection performance degrades for small objects on recovered noisy and low-resolution remote sensing images. Inspired by the success of edge enhanced GAN (EEGAN) and ESRGAN, we apply a new edge-enhanced super-resolution GAN (EESRGAN) to improve the image quality of remote sensing images and use different detector networks in an end-to-end manner where detector loss is backpropagated into the EESRGAN to improve the detection performance. We propose an architecture with three components: ESRGAN, Edge Enhancement Network (EEN), and Detection network. We use residual-in-residual dense blocks (RRDB) for both the ESRGAN and EEN, and for the detector network, we use the faster region-based convolutional network (FRCNN) (two-stage detector) and single-shot multi-box detector (SSD) (one stage detector). Extensive experiments on a public (car overhead with context) and a self-assembled (oil and gas storage tank) satellite dataset show superior performance of our method compared to the standalone state-of-the-art object detectors.

cs.CV↗

arXiv2019

MMGAN: Generative Adversarial Networks for Multi-Modal Distributions

Over the past years, Generative Adversarial Networks (GANs) have shown a remarkable generation performance especially in image synthesis. Unfortunately, they are also known for having an unstable training process and might loose parts of the data distribution for heterogeneous input data. In this paper, we propose a novel GAN extension for multi-modal distribution learning (MMGAN). In our approach, we model the latent space as a Gaussian mixture model with a number of clusters referring to the number of disconnected data manifolds in the observation space, and include a clustering network, which relates each data manifold to one Gaussian cluster. Thus, the training gets more stable. Moreover, MMGAN allows for clustering real data according to the learned data manifold in the latent space. By a series of benchmark experiments, we illustrate that MMGAN outperforms competitive state-of-the-art models in terms of clustering performance.

cs.LG↗

arXiv2018

Semi-Supervised Learning on Graphs Based on Local Label Distributions

Most approaches that tackle the problem of node classification consider nodes to be similar, if they have shared neighbors or are close to each other in the graph. Recent methods for attributed graphs additionally take attributes of neighboring nodes into account. We argue that the class labels of the neighbors bear important information and considering them helps to improve classification quality. Two nodes which are similar based on class labels in their neighborhood do not need to be close-by in the graph and may even belong to different connected components. In this work, we propose a novel approach for the semi-supervised node classification. Precisely, we propose a new node embedding which is based on the class labels in the local neighborhood of a node. We show that this is a different setting from attribute-based embeddings and thus, we propose a new method to learn label-based node embeddings which can mirror a variety of relations between the class labels of neighboring nodes. Our experimental evaluation demonstrates that our new methods can significantly improve the prediction quality on real world data sets.

cs.LG↗

arXiv2016

Scenic Routes Now: Efficiently Solving the Time-Dependent Arc Orienteering Problem

This paper extends the Arc Orienteering Problem (AOP) to large road networks with time-dependent travel times and time-dependent value gain, termed Twofold Time-Dependent AOP or 2TD-AOP for short. In its original definition, the NP-hard Orienteering Problem (OP) asks to find a path from a source to a destination maximizing the accumulated value while not exceeding a cost budget. Variations of the OP and AOP have many practical applications such as mobile crowdsourcing tasks (e.g., repairing and maintenance or dispatching field workers), diverse logistics problems (e.g., crowd control or controlling wildfires) as well as several tourist guidance problems (e.g., generating trip recommendations or navigating through theme parks). In the proposed 2TD-AOP, travel times and value functions are assumed to be time-dependent. The dynamic values model, for instance, varying rewards in crowdsourcing tasks or varying urgency levels in damage control tasks. We discuss this novel problem, prove the benefit of time-dependence empirically and present an efficient approximative solution, optimized for fast response systems. Our approach is the first time-dependent variant of the AOP to be evaluated on a large scale, fine-grained, real-world road network. We show that optimal solutions are infeasible and solutions to the static problem are often invalid. We propose an approximate dynamic programming solution which produces valid paths and is orders of magnitude faster than any optimal solution.

cs.DS↗

arXiv2016

End-to-end simulations of the Visible Tunable Filter for the Daniel K. Inouye Solar Telescope

The Visible Tunable Filter (VTF) is a narrowband tunable filter system for imaging spectroscopy and spectropolarimetry based. The instrument will be one of the first-light instruments of the Daniel K. Inouye Solar Telescope that is currently under construction on Maui (Hawaii). The VTF is being developed by the Kiepenheuer Institut fuer Sonnenphysik in Freiburg as a German contribution to the DKIST. We perform end-to-end simulations of spectropolarimetric observations with the VTF to verify the science requirements of the instrument. The instrument is simulated with two Etalons, and with a single Etalon. The clear aperture of the Etalons is 250 mm, corresponding to a field of view with a diameter of 60 arcsec in the sky (42,000 km on the Sun). To model the large-scale figure errors we employ low-order Zernike polynomials (power and spherical aberration) with amplitudes of 2.5 nm RMS. We use an ideal polarization modulator with equal modulation coefficients of 3-1/2 for the polarization modulation We synthesize Stokes profiles of two iron lines (630.15 nm and 630.25 nm) and for the 854.2 nm line of calcium, for a range of magnetic field values and for several inclination angles. We estimated the photon noise on the basis of the DKIST and VTF transmission values, the atmospheric transmission and the spectral flux from the Sun. For the Fe 630.25 nm line, we obtain a sensitivity of 20 G for the longitudinal component and for 150 G for the transverse component, in agreement with the science requirements for the VTF.

astro-ph.IM↗

arXiv2016

Skill-Based Differences in Spatio-Temporal Team Behavior in Defence of The Ancients 2

Multiplayer Online Battle Arena (MOBA) games are among the most played digital games in the world. In these games, teams of players fight against each other in arena environments, and the gameplay is focused on tactical combat. Mastering MOBAs requires extensive practice, as is exemplified in the popular MOBA Defence of the Ancients 2 (DotA 2). In this paper, we present three data-driven measures of spatio-temporal behavior in DotA 2: 1) Zone changes; 2) Distribution of team members and: 3) Time series clustering via a fuzzy approach. We present a method for obtaining accurate positional data from DotA 2. We investigate how behavior varies across these measures as a function of the skill level of teams, using four tiers from novice to professional players. Results indicate that spatio-temporal behavior of MOBA teams is related to team skill, with professional teams having smaller within-team distances and conducting more zone changes than amateur teams. The temporal distribution of the within-team distances of professional and high-skilled teams also generally follows patterns distinct from lower skill ranks.

stat.ML↗

arXiv2014

ParetoPrep: Fast computation of Path Skylines Queries

Computing cost optimal paths in network data is a very important task in many application areas like transportation networks, computer networks or social graphs. In many cases, the cost of an edge can be described by various cost criteria. For example, in a road network possible cost criteria are distance, time, ascent, energy consumption or toll fees. In such a multicriteria network, a route or path skyline query computes the set of all paths having pareto optimal costs, i.e. each result path is optimal for different user preferences. In this paper, we propose a new method for computing route skylines which significantly decreases processing time and memory consumption. Furthermore, our method does not rely on any precomputation or indexing method and thus, it is suitable for dynamically changing edge costs. Our experiments demonstrate that our method outperforms state of the art approaches and allows highly efficient path skyline computation without any preprocessing.

cs.DB↗

arXiv2011

Maximum Gain Round Trips with Cost Constraints

Searching for optimal ways in a network is an important task in multiple application areas such as social networks, co-citation graphs or road networks. In the majority of applications, each edge in a network is associated with a certain cost and an optimal way minimizes the cost while fulfilling a certain property, e.g connecting a start and a destination node. In this paper, we want to extend pure cost networks to so-called cost-gain networks. In this type of network, each edge is additionally associated with a certain gain. Thus, a way having a certain cost additionally provides a certain gain. In the following, we will discuss the problem of finding ways providing maximal gain while costing less than a certain budget. An application for this type of problem is the round trip problem of a traveler: Given a certain amount of time, which is the best round trip traversing the most scenic landscape or visiting the most important sights? In the following, we distinguish two cases of the problem. The first does not control any redundant edges and the second allows a more sophisticated handling of edges occurring more than once. To answer the maximum round trip queries on a given graph data set, we propose unidirectional and bidirectional search algorithms. Both types of algorithms are tested for the use case named above on real world spatial networks.

cs.SI↗