Brno University of Technology

About: Brno University of Technology is a education organization based out in Brno, Czechia. It is known for research contribution in the topics: Fracture mechanics & Filter (video). The organization has 6339 authors who have published 15226 publications receiving 194088 citations. The organization is also known as: Vysoké učení technické v Brně & BUT.

Design of the Human-Robot Interaction for a Semi-Autonomous Service Robot to Assist Elderly People

Abstract: Service robots could support elderly people’s activities of daily living and enable them to live in their own residences independently as long as possible. Current robot technology does not allow reliable fully autonomous operation of service robots with manipulation capabilities in the heterogeneous environments of private homes. We developed and evaluated a usage concept for semi-autonomous robot control as well as user interfaces for three user groups. Elderly people are provided with simple access to autonomous robot services through a handheld device. In case of problems with autonomous execution the robot contacts informal caregivers (e.g. relatives) who can support the robot using semi-autonomous teleoperation. To solve more complex problems, professional teleoperators are contacted who have extended remote access.

Gaze and conversational engagement in multiparty video conversation: an annotation scheme and classification of high and low levels of engagement

Abstract: When using a multiparty video mediated system, interacting participants assume a range of various roles and exhibit behaviors according to how engaged in the communication they are. In this paper we focus on estimation of conversational engagement from gaze signal. In particular, we present an annotation scheme for conversational engagement, a statistical analysis of gaze behavior across varying levels of engagement, and we classify vectors of computed eye tracking measures. The results show that in 74% of cases the level of engagement can be correctly classified into either high or low level. In addition, we describe the nuances of gaze during distinct levels of engagement.

Deep Clustering-Based Beamforming for Separation with Unknown Number of Sources.

Abstract: This paper extends a deep clustering algorithm for use with time-frequency masking-based beamforming and perform separation with an unknown number of sources. Deep clustering is a recently proposed single-channel source separation algorithm, which projects inputs into the embedding space and performs clustering in the embedding domain. In deep clustering, bi-directional long short-term memory (BLSTM) recurrent neural networks are trained to make embedding vectors orthogonal for different speakers and concurrent for the same speaker. Then, by clustering the embedding vectors at test time, we can estimate time-frequency masks for separation. In this paper, we extend the deep clustering algorithm to a multiple microphone setup and incorporate deep clustering-based timefrequency mask estimation into masking-based beamforming, which has been shown to be more effective than masking for automatic speech recognition. Moreover, we perform source counting by computing the rank of the covariance matrix of the embedding vectors. With our proposed approach, we can perform masking-based beamforming in a multiple-speaker case without knowing the number of speakers. Experimental results show that our proposed deep clustering-based beamformer achieves comparable source separation performance to that obtained with a complex Gaussian mixture model-based beamformer, which requires the number of sources in advance for mask estimation.

Deciding Entailments in Inductive Separation Logic with Tree Automata

Abstract: Separation Logic (SL) with inductive definitions is a natural formalism for specifying complex recursive data structures, used in compositional verification of programs manipulating such structures. The key ingredient of any automated verification procedure based on SL is the decidability of the entailment problem. In this work, we reduce the entailment problem for a non-trivial subset of SL describing trees (and beyond) to the language inclusion of tree automata (TA). Our reduction provides tight complexity bounds for the problem and shows that entailment in our fragment is EXPTIME-complete. For practical purposes, we leverage from recent advances in automata theory, such as inclusion checking for non-deterministic TA avoiding explicit determinization. We implemented our method and present promising preliminary experimental results.

Towards real-time intrusion detection for NetFlow and IPFIX

Abstract: DDoS attacks bring serious economic and technical damage to networks and enterprises. Timely detection and mitigation are therefore of great importance. However, when flow monitoring systems are used for intrusion detection, as it is often the case in campus, enterprise and backbone networks, timely data analysis is constrained by the architecture of NetFlow and IPFIX. In their current architecture, the analysis is performed after certain timeouts, which generally delays the intrusion detection for several minutes. This paper presents a functional extension for both NetFlow and IPFIX flow exporters, to allow for timely intrusion detection and mitigation of large flooding attacks. The contribution of this paper is threefold. First, we integrate a lightweight intrusion detection module into a flow exporter, which moves detection closer to the traffic observation point. Second, our approach mitigates attacks in near real-time by instructing firewalls to filter malicious traffic. Third, we filter flow data of malicious traffic to prevent flow collectors from overload. We validate our approach by means of a prototype that has been deployed on a backbone link of the Czech national research and education network CESNET.