Proceedings of the Workshop on Dynamic Networks Management and Mining, DyNetMM 2013
2013
Abstract:
For very large dynamic networks, monitoring the behavior of a subset of agents provides an efficient framework for detecting changes in network topology. For example, in mobile caller networks with millions of subscribers, we would like to monitor the dynamics of the smallest possible set of subscribers and still be able to infer abnormal events that occur over the entire network. In general, we assume that the temporal behavior of a network agent is captured by a (local) dynamic state, which may reflect either a physical property such as the number of connections or an abstract quantity such as opinions or beliefs. Further, assuming coupled linear inter-agent dynamics in which the local agent states evolve as weighted linear combinations of the neighboring agents’ states, we focus on tracking network-wide agent dynamics. Due to the large-scale nature of the problem, directly monitoring data streams of the state dynamics for every individual agent is infeasible. To address this issue, we propose a method that identifies a relatively small subset of agents whose state streams enable us to reconstruct the dynamic state evolution of all the network agents at any given time and, simultaneously, detect agent departure events. Using structural properties of the coupled inter-agent dynamics, we provide an algorithm, which is polynomial in the number of agents, to identify a small subset of agents that ensures such network observability regardless of any agent leaving. In addition, we show how well-known tools in dynamic control systems may be useful for identifying abnormal events; in particular, we use a fault detection and isolation scheme to identify agent departures. Finally, we illustrate our method and algorithms in a small test network as a proof of concept.
Sousa R. , Ferreira P., Costa P., Azevedo P., Costeira J.P., Santiago C., Magalhães J., Semedo D., Ferreira R., Rudnicky A., Hauptmann A.
Proceedings of the 2nd ACM Multimedia Workshop on Multimodal Conversational AI, MuCAI'21
2022
Abstract:
Most of the interaction between large organizations and their users will be mediated by AI agents in the near future. This perception is becoming undisputed as online shopping dominates entire market segments, and the new “digitally-native” generations become consumers. iFetch is a new generation of task-oriented conversational agents that interact with users seamlessly using verbal and visual information. Through the conversation, iFetch provides targeted advice and a “physical store-like” experience while maintaining user engagement. This context entails the following vital components: 1) highly complex memory models that keep track of the conversation, 2) extraction of key semantic features from language and images that reveal user intent, 3) generation of multimodal responses that will keep users engaged in the conversation and 4) an interrelated knowledge base of products from which to extract relevant product lists.
Diffusion of 3G cellular technology varies widely across countries and regions. Past studies have shown that lower levels of diffusion of previous technologies and higher levels of income are significant factors in accelerating the take up of 1st and 2nd generation of mobile telephony. In addition, spectrum management policy plays a significant role in shaping 3G diffusion. Regulatory policies regarding spectrum management include mandating band and technology and decisions to hold spectrum auctions. An econometric analysis over a multi-country panel dataset shows that these spectrum management policies do have significant influence on the take-up of 3G. Findings suggest that the presence of multiple technologies for the previous generation is associated with rollout delay. The estimations indicate that countries that mandated a specific frequency band for 3G saw faster roll out, but in the long run those countries experienced a slower growth rate. Also estimations find that 3G diffusion is not significantly affected by the choice of auctions vs. alternative license award processes. Insights gained from this study of the 2G to 3G transition can provide guidance to regulators now contemplating the transition to newer generations.
Boban M., Vinhoza T.T.V., Ferreira M., Barros J., Tonguz O.K.
IEEE Journal on Selected Areas in Communications
2011
Abstract:
A thorough understanding of the communications channel between vehicles is essential for realistic modeling of Vehicular Ad Hoc Networks (VANETs) and the development of related technology and applications. The impact of vehicles as obstacles on vehicle-to-vehicle (V2V) communication has been largely neglected in VANET research, especially in simulations. Useful models accounting for vehicles as obstacles must satisfy a number of requirements, most notably accurate positioning, realistic mobility patterns, realistic propagation characteristics, and manageable complexity. We present a model that satisfies all of these requirements. Vehicles are modeled as physical obstacles affecting the V2V communication. The proposed model accounts for vehicles as three-dimensional obstacles and takes into account their impact on the LOS obstruction, received signal power, and the packet reception rate. We utilize two real world highway datasets collected via stereoscopic aerial photography to test our proposed model, and we confirm the importance of modeling the effects of obstructing vehicles through experimental measurements. Our results show considerable obstruction of LOS due to vehicles. By obstructing the LOS, vehicles induce significant attenuation and packet loss. The algorithm behind the proposed model allows for computationally efficient implementation in VANET simulators. It is also shown that by modeling the vehicles as obstacles, significant realism can be added to existing simulators with clear implications on the design of upper layer protocols.
2013 IEEE Workshop on Automatic Speech Recognition and Understanding, ASRU 2013 - Proceedings
2013
Abstract:
In this paper, we present a technique to use the information in multiple parallel speech streams, which are approximate translations of each other, in order to improve performance in a punctuation recovery task. We first build a phraselevel alignment of these multiple streams, using phrase tables to link the phrase pairs together. The information so collected is then used to make it more likely that sentence units are equivalent across streams. We applied this technique to a number of simultaneously interpreted speeches of the European Parliament Committees, for the recovery of the full stop, in four different languages (English, Italian, Portuguese and Spanish). We observed an average improvement in SER of 37% when compared to an existing baseline, in Portuguese and English.
On-chip test/diagnosis is proposed to be an effective method to ensure the lifetime reliability of integrated systems. In order to manage the complexity of such an approach, an integrated system is partitioned into multiple modules where each module can be periodically tested, diagnosed and repaired if necessary. The limitation of on-chip memory and computing capability, coupled with the inherent uncertainty in diagnosis, causes the occurrence of misdiagnoses. To address this challenge, a novel incremental-learning algorithm, namely dynamic k-nearest-neighbor (DKNN), is developed to improve the accuracy of on-chip diagnosis. Different from the conventional KNN, DKNN employs online diagnosis data to update the learned classifier so that the classifier can keep evolving as new diagnosis data becomes available. Incorporating online diagnosis data enables tracking of the fault distribution and thus improves diagnostic accuracy. Experiments using various benchmark circuits (e.g., the cache controller from the OpenSPARC T2 processor design) demonstrate that diagnostic accuracy can be more than doubled.
ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
2013
Abstract:
We propose a method to combine audio of a lecture with its supporting slides in order to improve automatic speech recognition performance. We view both the lecture speech and the slides as parallel streams which contain redundant information. We integrate both streams in order to bias the recognizer’s language model towards the words in the slides, by first aligning the speech with the slide words, thus correcting errors on the ASR transcripts. We obtain a 5.9% relative WER improvement on a lecture test set, when compared to a speech recognition only system.
Monteiro R., Sargento S., Viriyasitavat W., Tonguz O.K.
IEEE Vehicular Networking Conference, VNC
2012
Abstract:
Developing routing protocols for Vehicular Ad Hoc Networks (VANETs) is a significant challenge in these large, self-organized and distributed networks. We address this challenge by studying VANETs from a network science perspective to develop solutions that act locally but influence the network performance globally. More specifically, we look at snapshots from highway and urban VANETs of different sizes and vehicle densities, and study parameters such as the node degree distribution, the clustering coefficient and the average shortest path length, in order to better understand the networks’ structure and compare it to structures commonly found in large real world networks such as small-world and scale-free networks. We then show how to use this information to improve existing VANET protocols. As an illustrative example, it is shown that, by adding new mechanisms that make use of this information, the overhead of the urban vehicular broadcasting (UV-CAST) protocol can be reduced substantially with no significant performance degradation.
We use cookies to ensure that we give you the best experience on our website. By using our website you agree to our Privacy Policy.
