Search

Data-driven deep learning solutions with gradient-based neural architecture, have proven useful in overcoming some limitations of traditional signal processing techniques. However, a large number of reverberant–anechoic training utterance pairs covering as many environmental conditions as possible is required to achieve robust dereverberation performance in unseen testing conditions. In this article, we propose to address the data requirement issue while preserving the advantages of deep neural structures leveraging upon hierarchical extreme learning machines (HELMs), which are not gradient-based neural architectures. In particular, an ensemble HELM learning framework is established to effectively recover anechoic speech from a reverberant one based on spectral mapping. In addition to the ensemble learning framework, we further derive two novel HELM models, namely, highway HELM [HELM(Hwy)] and residual HELM [HELM(Res)], both incorporating low-level features to enrich the information for spectral mapping. We evaluated the proposed ensemble learning framework using simulated and measured impulse responses by employing Texas Instrument and Massachusetts Institute of Technology (TIMIT), Mandarin hearing in noise test (MHINT), and reverberant voice enhancement and recognition benchmark (REVERB) corpora. The experimental results show that the proposed framework outperforms both traditional methods and a recently proposed integrated deep and ensemble learning algorithm in terms of standardized objective and subjective evaluations under matched and mismatched testing conditions for simulated and measured impulse responses.

The emerging wireless Network-on-Chip (WiNoC) architectures are a viable solution for addressing the scalability limitations of manycore architectures in which multi-hop long-range communications strongly impact both the performance and energy figures of the system. The energy consumption of wired links as well as that of radio communications account for a relevant fraction of the overall energy budget. In this article, we extend the approximate computing paradigm to the case of the on-chip communication system in manycore architectures. We present techniques, circuitries, and programming interfaces aimed at reducing the energy consumption of a WiNoC by exploiting the trade-off energy saving vs. application output degradation. The proposed platform—namely, xWiNoC—uses variable voltage swing links and tunable transmitting power wireless interfaces along with a programming interface that allows the programmer to specify those data structures that are error-resilient. Thus, communications induced by the access to such error-resilient data structures are carried out by using links and radio channels that are configured to work in a low energy mode, albeit by exposing a higher bit error rate. xWiNoC is assessed on a set of applications belonging to different domains in which the trade-off energy vs. performance vs. application result quality is discussed. We found that up to 50% of communication energy saving can be obtained with a negligible impact on the application output quality and 3% in application performance degradation.

The recent developments of information technologies, and the consequent need for access to distributed services and resources, require robust and reliable authentication systems. Biometric systems can guarantee high levels of security and multimodal techniques, which combine two or more biometric traits, warranting constraints that are more stringent during the access phases. This work proposes a novel multimodal biometric system based on iris and retina combination in the spatial domain. The proposed solution follows the alignment and recognition approach commonly adopted in computational linguistics and bioinformatics; in particular, features are extracted separately for iris and retina, and the fusion is obtained relying upon the comparison score via the Levenshtein distance. We evaluated our approach by testing several combinations of publicly available biometric databases, namely one for retina images and three for iris images. To provide comprehensive results, detection error trade‐off‐based metrics, as well as statistical analyses for assessing the authentication performance, were considered. The best achieved False Acceptation Rate and False Rejection Rate indices were and 3.33%, respectively, for the multimodal retina‐iris biometric approach that overall outperformed the unimodal systems. These results draw the potential of the proposed approach as a multimodal authentication framework using multiple static biometric traits.

Model adaptation is a key technique that enables a modern automatic speech recognition (ASR) system to adjust its parameters, using a small amount of enrolment data, to the nuances in the speech spectrum due to microphone mismatch in the training and test data. In this brief, we investigate four different adaptation schemes for connectionist (also known as hybrid ) ASR systems that learn microphone-specific hidden unit contributions, given some adaptation material. This solution is made possible adopting one of the following schemes: 1) the use of Hermite activation functions; 2) the introduction of bias and slope parameters in the sigmoid activation functions; 3) the injection of an amplitude parameter specific for each sigmoid unit; or 4) the combination of 2) and 3). Such a simple yet effective solution allows the adapted model to be stored in a small-sized storage space, a highly desirable property of adaptation algorithms for deep neural networks that are suitable for large-scale online deployment. Experimental results indicate that the investigated approaches reduce word error rates on the standard Spoke 6 task of the Wall Street Journal corpus compared with unadapted ASR systems. Moreover, the proposed adaptation schemes all perform better than simple multicondition training and comparable favorably against conventional linear regression-based approaches while using up to 15 orders of magnitude fewer parameters. The proposed adaptation strategies are also effective when a single adaptation sentence is available.

This paper introduces a fuzzy-based method that, according to the ratio of Throughput to Workload and the battery level, manages the sleeping time of devices in Wireless Sensor Networks (WSNs) for smart homes. The purpose of this work is a system that can be executed on off-the-shelf hardware and offers enhanced performance confronted with other approaches. The challenge here is to achieve a practical method that reaches the target while bypassing complex and computationally expensive solutions, which would diminish the possible applicability of the method in real scenarios. The retrieved results prove that the proposed approach outperforms other solutions, significantly prolonging the life of battery-powered wireless devices with also satisfactory values of the ratio Throughput to Workload. Besides, a proof-of-concept implementation on off-the-shelf devices confirms that the proposed method does not expect powerful hardware and can be surely implemented on a low-cost device.

Smart homes represent one of the principal points in the new ecosystem of the Internet of Things (IoT), both for the centrality of the home in the life of individuals and the significant potential concerning the diffusion of smart objects and innovative services. While IoT-oriented smart homes can revise how inhabitants interact with the domestic environment, each well-defined piece of technology necessitates precise network performance and distinct levels of security based on the sensitivity of the controlled system and the information it handles. This editorial presents a review of the papers accepted in the special issue. The issue has focused at obtaining high-quality papers aimed at solving well-known technical problems and challenges typical of IoT-oriented smart homes.

Power disaggregation is aimed at determining appliance-by-appliance electricity consumption, leveraging upon a single meter only, which measures the entire power demand. Data-driven procedures based on Factorial Hidden Markov Models (FHMMs) have produced remarkable results on energy disaggregation. Nevertheless, these procedures have various weaknesses, there is a scalability problem as the number of devices to observe rises, and the inference step is computationally heavy. Artificial neural networks (ANNs) have been demonstrated to be a viable solution to deal with FHMM shortcomings. Nonetheless, there are two significant limitations: A complicated and time-consuming training system based on back-propagation has to be employed to estimate the neural architecture parameters, and large amounts of training data covering as many operation conditions as possible need to be collected to attain top performances. In this work, we aim to overcome these limitations by leveraging upon the unique and useful characteristics of the extreme learning machine technique, which is based on a collection of randomly chosen hidden units and analytically defined output weights. We find that the suggested approach outperforms state-of-the-art solutions, namely FHMMs and ANNs, on the UK-DALE corpus. Moreover, our solution generalizes better than previous approaches for unseen houses, and avoids a data-hungry training scheme.

The aim of this paper is to discuss the use deep neural networks (DNNs) for road detection from high-resolution aerial images. DNNs are a powerful mathematical model for information processing that has been proven to be effective to accomplish several complex tasks including coding and classification of image data. The DNN with its many trainable weights allows taking a large context into account that is essential for the difficult problem of detect roads from aerial image. Furthermore it allows to progressively combining lower-level features or concepts into more abstract and higher-level representations. Thus, we believe useful to discuss possible applications DNNs for roads detection in the present work.

The increasing of cars in our planet and the consequent congestion of urban and intercity traffic, opens new challenges in ITS (Intelligent Transportation System). They can refer to several aspects. This paper analyses Carpooling, a technique of private cars sharing diffused in United Kingdom, United States and Italy to mention some. It helps to handle several troubles, such as the release of CO2 in the air, noise pollution, and excessive fuel consumption. Carpooling is considered an efficient technique, even if some relevant aspects should be improved. First of all, to have real time notifications about lift possibilities. In addition, a mechanism to increase trust should be introduced: people hardly offer a lift to strangers. This work shows a trust-based algorithm to encourage people interested to carpooling as alternative way to move themselves inside or outside cities. The proposed algorithm, that assigns priority to users with positive feedbacks through a fuzzy logic scheme, has the specific goal to increase the level of trust, decreasing significantly the possibility that, during a trip, unpleasant episodes occur.

Advantages provided by the use of wireless technologies in industrial scenarios, like reduced costs of installation, better flexibility, the ability to easily realize temporary deployments (e.g. for monitoring purposes), etc. are discussed by researchers in literature. In these contexts, time-critical transmissions have to be supported by automation networks. In IEEE 802.11 standard protocol, one of most popular technologies for process control networks, Access Points (APs) are responsible for nodes connection. Timeliness of communications can be compromised by transmission of wireless nodes through overloaded APs. This paper shows a new dynamic Load Balancing approach that distributes network load in order to ensure, for each station, best possible performances in terms of soft real-time constraints, in an industrial process control scenario.

In this chapter an approach to improve road monitoring through wireless sensor networks is described: several sensors may be distributed along a road in order to detect traffic flows, speed, and the continued occupation of the road. Each WSN cluster offers a monitoring system based on videotapes and, at the same time, produces information which can be processed to provide a clear interpretation of the road situation regardless of weather conditions [1, 2]. Therefore, it is necessary to integrate video monitoring information with parameters measured by other sensors, e.g. magnetometer or power sensors, for traffic detection [3–5]. Advantages about WSN usage in this area are several. Some of them have been already shown in literature. Others derive by the proposed approach.

This paper shows an innovative Wireless Sensor Network (WSN) architecture based on a flexible Quality of Service (QoS) approach for road traffic management. The aim of this work is to analyse an algorithm that dynamically enables/disables some cameras according to the real need to monitor a given area, It is based on traffic volume measured values. This approach has been developed in order to manage both network topology and workload conditions, using a fuzzy logic controller for a flexible QoS management. Performances of this approach have been evaluated using True Time and Simulink/Matlab.

The network management plays an important role in Wireless Sensor Networks WSNs, to control both the sleep and the wake times of the nodes to make data transmissions in an energy-efficient way, thus prolonging network lifetime. In this paper, we propose an innovative approach to control the uptime of the nodes of an Industrial WSN using a fuzzy based algorithm. In a scenario of process control is not possible to separate power energy management from the need to ensure respect for real-time constraints imposed by the application such as deadline miss values of periodic traffic flows. The mechanism here described takes into account the timing requirements of the soft real-time application and, at the same time, minimizes energy consumption. Therefore the aim of this paper is to show how, using a fuzzy logic controller, implemented in an industrial WSN, it is possible increasing the life cycle of the batteries of the individual wireless nodes and by exerting continuous control on energy consumption of individual nodes and on network performance, by adapting in a dynamic way the sampling period of traffic flows of the WSN. To show the effectiveness of our approach, we also evaluated its performance in True-Time.

We have recently proposed a universal acoustic characterisation to foreign accent recognition, in which any spoken foreign accent was described in terms of a common set of fundamental speech attributes. Although experimental evidence demonstrated the feasibility of our approach, we belive that speech attributes, namely manner and place of articulation, can be better modelled by a deep neural network. In this work, we propose the use of deep neural network trained on telephone bandwidth material from different languages to improve the proposed universal acoustic characterisation. We demonstrate that deeper neural architectures enhance the attribute classification accuracy. Furthermore, we show that improvements in attribute classification carry over to foreign accent recognition by producing a 21% relative improvement over previous baseline on spoken Finnish, and a 5.8% relative improvement on spoken English. Index Terms: Deep neural networks, data-driven speech attributes, manner of articulation, place of articulation, i-Vector, foreign accent recognition