Alle Publikationen
- <<
- <
- 1
2016
-
(2016): Semantic language models with deep neural networks. In: COMPUTER SPEECH AND LANGUAGE 40, S. 1-22
Abstract: In this paper we explore the use of semantics in training language models for automatic speech recognition and spoken language understanding. Traditional language models (LMs) do not consider the semantic constraints and train models based on fixed-sized word histories. The theory of frame semantics analyzes word meanings and their constructs by using "semantic frames". Semantic frames represent a linguistic scene with its relevant participants and their relations. They are triggered by target words and include slots which are filled by frame elements. We present semantic LMs (SELMs), which use recurrent neural network architectures and the linguistic scene of frame semantics as context. SELMs incorporate semantic features which are extracted from semantic frames and target words. In this way, long-range and "latent" dependencies, i.e. the implicit semantic dependencies between words, are incorporated into LMs. This is crucial especially when the main aim of spoken language systems is understanding what the user means. Semantic features consist of low-level features, where frame and target information is directly used; and deep semantic encodings, where deep autoencoders are used to extract semantic features. We evaluate the performance of SELMs on publicly available corpora: the Wall Street Journal read-speech corpus and the LUNA human-human conversational corpus. The encoding of semantic frames into SELMs improves the word recognition performance and especially the recognition performance of the target words, the meaning bearing elements of semantic frames. We assess the performance of SELMs for the understanding tasks and we show that SELMs yield better semantic frame identification performance compared to recurrent neural network LMs.
-
(2016) : Modularization of skill ontologies for industrial robots In: VDE: Proceedings of ISR 2016: 47st International Symposium on Robotics: Berlin: VDE Verlag GmbH, S. 1-6. Online verfügbar unter https://ieeexplore.ieee.org/document/7559114/
Abstract: With industrial robots ready to take the next step in mastering manufacturing tasks new approaches to reduce the programming effort are needed. This is achieved by introducing skills as robot "know-how" and using them as a higher abstraction level of robot instructions during programming. The skills are reusable items providing motion control and rich declarative descriptions of complex robot capabilities. Storing the skills requires an adequate knowledge representation model that enables reuse and reasoning on skills and simplifies knowledge management. In this paper we report on development of a skill representation model and its implementation in a knowledge base. The developed model is effectively a class hierarchy of the skill concepts implemented in a modularized ontology structure. The resulting model clarifies the intrinsic concepts of a skill and presents a module structure that enables the future development and reuse of skills in general.
2007
-
(2007): A Dual Mechanism Neural Framework for Social Understanding. In: Philosophical Psychology 20 (1), S. 43-63. DOI: 10.1080/09515080601085864
Abstract: In this paper a theoretical framework is proposed for how the brain processes the information necessary for us to achieve the understanding of others that we experience in our social worlds. Our framework attempts to expand several previous approaches to more fully account for the various data on interpersonal understanding and to respond to theoretical critiques in this area. Specifically, we propose that social understanding must be achieved by at least two mechanisms in the brain that are capable of parallel information processing. The first mechanism, based on research into mirror matching systems in the brain, suggests that representations of others are mapped onto an observer's representations of these same schemas in order to understand them. The second mechanism requires semantic analysis of a given social situation in order to understand the actions of others and most likely involves conscious processes. We suggest that experimental correlates of these systems should be dissociable using both behavioral and neuroimaging techniques.
Keywords: Gehirnfunktion, Intellektualtechnik, Interaktionspartner, interpersonal, interpersonal communication, Kogn. Architektur, Kognitionswissenschaft/Social Sciences/Humanities, kognitive Architekturen, Kognitive Skills/Social Cognition, Künstliche Intelligenz, Mensch-Technik-Relationen (MTR), Neuronale Netzwerke, Neurowissenschaften, Philosophical Psychology, Realtechnik, Semantics, Semantik, Situationsbedingter Kontext, Sozialverstehen, Technik, theory of mind 1997
-
(1997): Long Short-Term Memory. In: Neural Computation 9 (8), S. 1735-1780. DOI: 10.1162/neco.1997.9.8.1735
DOI: https://doi.org/10.1162/neco.1997.9.8.1735 Abstract: Learning to store information over extended time intervals via recurrent backpropagation takes a very long time, mostly due to insucient, decaying error back ow. We brie y review Hochreiter's 1991 analysis of this problem, then address it by introducing a novel, ecient, gradient-based method called \Long Short-Term Memory" (LSTM). Truncating the gradient where this does not do harm, LSTM can learn to bridge minimal time lags in excess of 1000 discrete time steps by enforcing constant error ow through \constant error carrousels" within special units. Multiplicative gate units learn to open and close access to the constant error ow. LSTM is local in space and time; its computational complexity per time step and weight is O(1). Our experiments with articial data involve local, distributed, real-valued, and noisy pattern representations. In comparisons with RTRL, BPTT, Recurrent Cascade-Correlation, Elman nets, and Neural Sequence Chunking, LSTM leads to many more successful runs, and learns much faster. LSTM also solves complex, articial long time lag tasks that have never been solved by previous recurrent network algorithms.
Keywords: deutsche Community, Intellektualtechnik, Kogn. Architektur, kognitive Architekturen, Künstliche Intelligenz, Mensch-Technik-Relationen (MTR), Neuronale Netzwerke, Programmieren, Realtechnik, Sprachverstehen, Technik, Wiedererkennen / Erinnerung / Erkennen von Personen und Sitautionen, Social Recognition
- <<
- <
- 1