Search our Publications
Latest Publications
Heiko Wersing, Stephan Hasler, Sebastian Schrom, Lydia Fischer , "Improving online learning of visual categories by deep features", Workshop Future of Interactive Learning Machines (FILM) at NIPS 2016, 2016.
AbstractWe review recent work on incremental architectures for real-time object category learning in robotics. We train a deep architecture for a 126 object identification task with free fully rotated and unconstrained in-hand presentation. We show that the feature representation that was trained for object identification is highly efficient for learning more abstract shape and color categories. The resulting performance strongly exceed our earlier resul...
Download Bibtex file
Viktor Losing, Barbara Hammer, Heiko Wersing , "Dedicated Memory Models for Continual Learning in the Presence of Concept Drift", Continual Learning and Deep Networks Workshop NIPS 2016, 2016.
AbstractLearning in non-stationary data streams is is a highly challenging task, since the fundamentally different types of possibly occurring drift undermine classical assumptions such as i.i.d. data or stationary distributions. Available algorithms are either struggling with certain forms of drift or require a priori knowledge in terms of a task specific setting. We propose a method which couples the KNN classifier with dedicated memories for the...
Download Bibtex file
Nadja Schömig, Martin Ernst Heckmann, Heiko Wersing, Christian Maag, Alexandra Neukum , "“Assistance-on-demand”: A speech-based assistance system for urban intersections", Adjunct Proc. AutomotiveUI, 2016.
AbstractWe evaluated a system to support the driver in urban intersections (called “Assistance on Demand” AoD system). The system is controlled via speech and supports the driver in monitoring and decision making by providing recommendations for suitable time gaps to enter the intersection. This speech-based control of the system allows the implementation of an „on-demand“-concept where the driver can activate the assistance only if he desires support. 2...
Download Bibtex file
Andrea Schnall and Martin Ernst Heckmann , "COMPARING SPEAKER INDEPENDENT AND SPEAKER ADAPTED CLASSIFICATION FOR WORD PROMINENCE DETECTION", 2016 IEEE Workshop on Spoken Language Technology, 2016.
AbstractProminence is an essential tool in human communication to e.g. express important information. But since there are large differences how prominence is expressed by different speakers, it is not easy to extract and to incorporate it in a speech recognition system. In speech processing, for the handling of new data in a speaker independent trained model, adaptation techniques has been established. Most speaker adaptation techniques are developed fo...
Download Bibtex file
Ran Cheng, Yaochu Jin, Markus Olhofer, Bernhard Sendhoff , "Test Problems for Large-Scale Multi- and Many-Objective Optimization", IEEE Transactions on Cybernetics, vol. 47, no. 12, pp. 4108-4121, 2016.
AbstractThe interests in multi- and many-objective optimization have been rapidly increasing in the evolutionary computation community. However, in the literature, the majority of the study on multi- and many-objective optimization is limited to small-scale decision variables, notwithstanding real-world multi- and many-objective optimization problems may involve large-scale decision variables as well. One factor that limits the development of large-scal...
Download Bibtex file
Marvin Struwe, Stephan Hasler, Ute Bauer-Wersing , "Rendered Benchmark Data Set for Evaluation of Occlusion-handling Strategies of a Parts-based Car Detector", Pacific Rim Symposium on Image and Video Technology, 2015.
AbstractIn this paper we extend a parts-based detection model with a simple occlusion handling strategy. We test it on a rendered car data set which allows us a more systematic evaluation of the influence of occlusion....
Download Bibtex file
Irene Ayllón Clemente , "Towards natural speech acquisition: incremental word learning with limited data", University of Bielefeld, 2015.
AbstractA strong trend in robotics is the investigation of adaptable machine learning algorithms and frameworks that enhance the skills and application of artificial systems during the interaction with humans. The use of language is one of the most convenient human methods to communicate with artificial agents. In the last decades, the introduction of automatic speech recognition (ASR) systems achieved important advances in the field, ...
Download Bibtex file Per Mail Request
Julian Eggert, Stefan Klingelschmitt, Florian Damerow , "The Foresighted Driver: Future ADAS Based on Generalized Predictive Risk Estimation", FAST-zero 2015 Symposium, issue 3rd International Symposium on Future Active Safety Technology Towards zero traffic accidents, pp. 93-100, 2015.
AbstractSeparably developed functionality as well as increasing situation complexity poses problems for building, testing, and validating future ADAS. These will have to deal with situations in which several current ADAS domains interplay. We argue that a generalized estimation of the future ADAS functions benefit is required for efficient testing and evaluations, and propose a quantification based on an estimation of the predicted risk. We show that suc...
Download Bibtex file
Stefan Klingelschmitt and Julian Eggert , "Using Context Information and Probabilistic Classification for Making Extended Long-Term Trajectory Predictions", 18th IEEE International Conference on Intelligent Transportation Systems 2015 (ITSC), pp. 705-711, 2015.
AbstractIntersections are among the most accident prone spots in traffic. Future Advanced Driver Assistance Systems (ADAS) are aiming to assist the driving task in these complex scenarios. This can be realized by assessing the criticality of possible occurring situations. For such criticality assessment techniques predicting the trajectories of the involved traffic participants several seconds in advance are required. In this paper we outline a method th...
Download Bibtex file
Julian Eggert and Florian Damerow , "Complex Lane Change Behavior in the Foresighted Driver Model", 18th IEEE International Conference on Intelligent Transportation Systems 2015 (ITSC), pp. 1747 - 1754, 2015.
AbstractThis paper presents a general model for lane change decision and dynamic lane change behavior based on the Foresighted Driver Model. Differently to previous models, it does not rely on indirect features like gap lengths, TTC or decelerations, but it models the lane change process by explicitly calculating the involved risks and benefits of the different trajectory alternatives. The lane change decision occurs by searching a tradeoff between the a...
Download Bibtex file
1 ... 96 97 98 99 100 101 ... 183
Follow us
