BME MDA - a Műegyetem Digitális Archivuma
Not a member yet
47877 research outputs found
Sort by
Ethics of AI
No full textThis entry presents a comprehensive overview of AI ethics, delving into various ethical and social issues arising from AI's expanding role in different facets of private and societal life. It sheds light on critical challenges such as accountability gaps, biases, discrimination, and risks associated with generative AI technologies which are relevant for information flows and the public communication system. The entry then explores the emergence of concepts like Trustworthy AI and Responsible AI as responses to these challenges, underscoring their importance in developing AI systems that minimize risks and maximize societal benefits. Despite the growing optimism around Trustworthy AI, the entry also emphasizes the difficulties in implementing ethical principles within AI systems. This opens the discussion to critical perspectives that, while not mainstream in AI ethics, offer valuable insights into power dynamics inherent in AI technologies. Furthermore, the article addresses the ongoing debate on AI regulation and standardization, framing it as an extension of the initial AI ethics discourse. Lastly, it outlines the relevance of analyzing public and political discourse on AI to scholars and practitioners in the field of political communication
Conformer-based Automatic Speech Recognition for Arabic Dialects
No full textAutomatic Speech Recognition has shown a significant upward trend in recent years. This paper investigates an ASR system for the Arabic language, developed using the Conformer-CTC character-based model within the NeMo framework. The system leverages the latest deep learning techniques, focusing on the conformer architecture combined with Connectionist Temporal Classification for sequence-to-sequence learning. The model is supervised, using labeled training data to map the input audio to text. The Mozilla Common Voice 11.0 dataset, which offers diverse spoken Arabic samples, is used for training. This paper details the model training process, including configuration setup, data processing, and optimization strategies. The performance of the model is evaluated, offering insights into the challenges and effectiveness of the Conformer-CTC character-based model for Arabic speech recognition tasks
Integrált szoftverkörnyezet fejlesztése valós idejű CAN forgalom monitorozásához és szimulálásához
No full text
Towards Configurable Coordination for Distributed Reactive Systems
Full text linkModern cyber-physical systems (CPS) present unique challenges as they are distributed real-time systems used in many critical application domains, such as automotive or railway systems. However, modeling and verifying the distributed and timed aspects of the system are challenging tasks. We need a precise description of the possible orderings of the components' execution, and a formal representation to be able to run formal verification. In this paper, we investigate the possible extension of a modeling and verification framework to support the flexible, configurable description of the coordination of distributed critical systems. We present an extension of the timed automata formalism, the coordination automata formalism, and show the applicability of the extension on a motivating example
