Search

A recent trend in machine learning has been to enrich learned models with the ability to explain their own predictions. The emerging field of Explainable AI (XAI) has so far mainly focused on supervised learning, in particular, deep neural network classifiers. In many practical problems however, label information is not given and the goal is instead to discover the underlying structure of the data, for example, its clusters. While powerful methods exist for extracting the cluster structure in data, they typically do not answer the question why a certain data point has been assigned to a given cluster. We propose a new framework that can, for the first time, explain cluster assignments in terms of input features in an efficient and reliable manner. It is based on the novel insight that clustering models can be rewritten as neural networks - or 'neuralized'. Cluster predictions of the obtained networks can then be quickly and accurately attributed to the input features. Several showcases demonstrate the ability of our method to assess the quality of learned clusters and to extract novel insights from the analyzed data and representations., 15 pages + supplement

Globally, disasters disrupt human lives and women tend to be more vulnerable during such events. This narrative review explores women’s experiences during disasters and identifies common factors increasing their vulnerability. After critical reading, 39 articles were included in this review. This paper underlines the themes in that literature to show that women across the world experience domestic violence, sexual assault, psychological and health problems as well as social and financial deprivation in disasters. The paper discusses the vulnerability of women particularly in Australia and New Zealand, through the lens of the global experience of women in disaster. This review highlights that, while there is consensus on the challenges faced by women in Australia, more research regarding interventions is required to reduce the negative effects of disasters on women. This review aims to inform emergency management practice in Australia and to direct further research to improve the outcomes for women and their safety.

Domain translation is the task of finding correspondence between two domains. Several deep neural network (DNN) models, e.g., CycleGAN and cross-lingual language models, have shown remarkable successes on this task under the unsupervised setting--the mappings between the domains are learned from two independent sets of training data in both domains (without paired samples). However, those methods typically do not perform well on a significant proportion of test samples. In this article, we hypothesize that many of such unsuccessful samples lie at the fringe--relatively low-density areas--of data distribution, where the DNN was not trained very well, and propose to perform the Langevin dynamics to bring such fringe samples toward high-density areas. We demonstrate qualitatively and quantitatively that our strategy, called Langevin cooling (L-Cool), enhances state-of-the-art methods in image translation and language translation tasks.

The potential of broadband THz communication system is not limited to extremely high data rate transmissions up to several TBit/s. Rather, the true application potential of THz systems will be created by the combination of the communication and radar/sensor functionality under the use of the enormous available bandwidth with partially more than 10 GHz. This enables to develop new applications in the areas of 3D imaging radar systems, sensor systems, car‐to‐car communication in combination with car radar, and security application like private alarm system via Wi‐Fi router. The reuse of future standard 3GPP and IEEE802 THz technology will allow the implementation of low‐cost THz communication systems, which enable the joint use of THz systems as a high‐resolution sensor systems, which optimize themselves by the use of artificial intelligence to provide never‐before‐achieved resolutions for radar and sensor applications additional to high‐speed Tbit/s data transmissions.

On dit et repete que, dans la Republique, Platon a chasse les poetes de la cite. L’affirmation n’est pas fausse, a condition d’ajouter plusieurs reserves et precisions qui, si on les prend au serieux, donnent une image tres differente de l’attitude de Platon envers la poesie. La cite a besoin de poetes : une partie essentielle de l’education des gardiens (livres II-III) repose sur la musique-poesie, et c’est bien pourquoi Platon s’attarde longuement sur les regles a respecter en la matiere. Cette poesie soumise a la censure n’est pourtant pas la seule admise : vis-a-vis de la poesie existante (Homere, les Tragiques), le philosophe adopte une position nuancee : elle a beaucoup de charme, et si elle a beaucoup de defauts, il n’est pourtant pas impossible de l’accueillir dans la cite pour peu que les citoyens soient munis de l’antidote fourni par l’education. C’est a cette derniere qu’on doit apporter tous ses soins, a celle des gardiens mais aussi a celle des citoyens ordinaires.

Bangladesh is repeatedly threatened by tropical storms and cyclones, exposing one-third of the total population of the country. As a preparedness measure, several cyclone shelters have been constructed, yet a large proportion of the coastal population, especially women, are unwilling to use them. Existing studies have demonstrated a range of concerns that discourage women from evacuating and have explored the limitations of the shelters, but the experiences of female evacuees have not been apparent in these stories. This study explores the lived-experiences of women in the cyclone shelters of Bangladesh and discusses their health and well-being as evacuees in the shelters. Nineteen women from three extremely vulnerable districts of coastal Bangladesh were interviewed. Seven research themes were identified from the participants’ narratives using van Manen’s thematic analysis process. The most salient theme, being understood (as a woman), portrayed the quintessential image of these women, which subsequently influenced their vulnerability as evacuees. The next themes–being a woman during crisis, being in a hostile situation, being fearful, being uncertain, being faithful, and being against the odds–focused on the incidents they lived through which affected their physical and mental health and the emotions they felt as evacuees. The paper offers a deep inquiry into women’s experiences of well-being in the shelters and recognizes the significance of women’s voices to improve their experiences as evacuees.

Due to the great demand of throughput and reliability for multimedia applications in Fifth Generation (5G) networks, many broadcasting systems adopt the Millimeter Wave (mmWave) technology to address the lack of the spectrum resources. As one of 5G-PPP projects, Internet of Radio Light (IoRL) project adopts 40GHz mmWave band to support a high-speed and stable Ultra-High-Definition (UHD) television broadcasting service in the indoor environment. Because of the high frequency property, mmWave bands usually suffers from the high path loss and the penetration loss. Thus, in order to overcome these issues, directional antennas are employed to provide additional power gain while increasing transmission distance. However, the mmWave with directional antennas brings additional problems, such as limited transmission angle and more multipath effects. Therefore, in this paper, for better understanding of impact factors on the signal quality and transmission coverage of the directional 40GHz mmWave band in the indoor environment, a measurement campaign is introduced in detail and the channel characteristics are measured and analysed in varying cases. The mainly concerned characteristics are path loss, shadow fading, average Power Delay Profile (PDP), Root-Mean-Square (RMS) delay spread, arrival rate and coherence bandwidth. All Measured characteristic values are summarised in three tables at the end of this paper. Besides of these, as a reference of channel analysis and a metric of signal quality and effective coverage, Error Vector Magnitude (EVM) of received signal in each case is measured and discussed. Moreover, a simulation is performed based on a statistical channel model to validate the measured results.

BACKGROUND The internet has become an increasingly popular medium for parents to obtain health information. More studies investigating the impact of paid digital marketing campaigns for parents on promoting children’s healthy development are needed. OBJECTIVE This study aims to explore the outcomes of a paid digital marketing campaign, which occurred from 2018 to 2020, to promote messages about parent-engaged developmental monitoring and ultimately direct parents to the Centers for Disease Control and Prevention’s (CDC’s) Milestone Tracker app, a mobile health (mHealth) app developed by the CDC. METHODS The paid digital marketing campaign occurred in 3 phases from 2018 to 2020. In each phase, 24 to 36 marketing messages, in English and Spanish, were created and disseminated using Google’s Universal App Campaigns and Facebook Ads Manager. Outcomes were measured using impressions, clicks, and install data. Return on investment was measured using click-through rate (CTR), cost per click, and cost per install metrics. RESULTS The Google-driven marketing messages garnered a total of 4,879,722 impressions (n=1,991,250, 40.81% for English and n=2,888,472, 59.19% for Spanish). The messages resulted in a total of 73,956 clicks (n=44,328, 59.94% for English and n=29,628, 40.06% for Spanish), with a total average CTR of 1.52% (2.22% for English and 1.03% for Spanish). From these clicks, there were 13,707 installs (n=9765, 71.24% for English and n=3942, 28.76% for Spanish) of the CDC’s Milestone Tracker app on Google Play Store. The total average cost per install was US $0.93 across all phases. The phase 3 headline “Track your child’s development” generated the highest CTR of 3.23% for both English and Spanish audiences. The Facebook-driven marketing messages garnered 2,434,320 impressions (n=1,612,934, 66.26% for English and n=821,386, 33.74% for Spanish). The messages resulted in 44,698 clicks (n=33,353, 74.62% for English and n=11,345, 25.38% for Spanish), with an average CTR of 1.84% (2.07% for English and 1.38% for Spanish). In all 3 phases, animated graphics generated the greatest number of clicks among both English and Spanish audiences on Facebook when compared with other types of images. CONCLUSIONS These paid digital marketing campaigns can increase targeted message exposure about parent-engaged developmental monitoring and direct a parent audience to an mHealth app. Digital marketing platforms provide helpful metrics that can be used to assess the reach, engagement, and cost-effectiveness of this effort. The results from this study suggest that paid digital marketing can be an effective strategy and can inform future digital marketing activities to promote mHealth apps targeting parents of young children.

The lower THz bands are foreseen as candidates to achieve the data-rates demanded in the sixth generation (6G) of wireless communications. Hence, multiple measurements for characterization of propagation at these frequencies in different scenarios are being conducted all around the world. This also impulses activities on standardization of measurement equipment and methodologies to define common practices and make measurements from different actors comparable. Therefore, in the present paper we investigate the accuracy of a dual-polarized ultra-wideband channel sounder in the time, angular, and polarization domains by means of a verification methodology based on interferometry.

The second-order statistics of indoor directional channels are investigated using millimeter-wave (mmWave) band (30.4–37.1 GHz) ultrawideband (UWB) channel measurements. Considering two main mmWave system assumptions (high bandwidth and high beamforming gain), this paper aims to investigate the validity of the Rayleigh–Rice fading models for the cluster fading envelope. The results from the mmWave band study are compared to an already well-studied lower frequency FCC band (3.4–10.1 GHz). During the measurements, only selective objects (emulated multipath clusters in the propagation channel) are illuminated in a small lecture room. The experiments show that for both UWB channels, the complex received (Rx) signal is a circularly symmetric non-Gaussian random variable with highly correlated inphase (I) and quadrature (Q) components. These properties demonstrate that the intracluster multipath components (MPCs) structure is sparse. Consequently, modeling the cluster fading envelope with Rayleigh–Rice distribution is not realistic. Therefore, the sum-of-cisoids principle is used for intracluster multipath modeling which inherently considers a correlation between I and Q components. It has been established that a reasonably good approximation of the cluster fading envelope can be obtained with $N=3-6$ equal amplitude cisoids. However, we remark that Rayleigh–Rice models may become realistic cluster fading envelopes for narrowband mmWave systems.

The characterization of propagation at mm-waves and THz is obtaining relevance since they are expected to be the frequency bands of the future wireless generations. Hence, measurements conducted all around the world with different equipments need to be comparable by establishing common methodologies and best-practice guidelines. As an example, all measurements are affected by noise, but its estimation and removal are not standardized. Therefore, in the present paper we define a methodology on the estimation of the noise floor and the cut-off margin for its subsequent removal. We firstly analyse the influence of the channel sounder noise floor on the estimation of different propagation parameters directly from the measured channel impulse responses (CIRs), showing the need of its removal before any processing of the measurements. Later, we introduce an algorithm to estimate the noise floor in the presence of signal, and the cut-off margin. Finally, we evaluate its performance empirically with measurements.

Phytochemical study of

As one of the key features in 5G network, Millimeter wave (mmWave) technology can provide the ultra-wide bandwidth to support higher data rate. However, for high frequency band, mmWave signal still suffers from the high pathloss, the multipath fading and the signal blockage issue, especially in the indoor environment. For different application scenarios, the channel conditions and quality of services (QoS) are quite different. Therefore, it is essential to investigate the impact of the mmWave channel on the system performance. This paper investigates and measures the performance of a 60GHz mmWave module that is exploited for the downlink and uplink high data rate transmission in the Internet of Radio-Light (IoRL) project. The coverage area and the throughput of the mmWave module is estimated by measuring the error vector magnitude (EVM) of received signals with different transmitter (TX) and receiver (RX) angles and at different locations in a laboratory. In this paper, the measurement environment and system setup are introduced. After that, the waveform design for the measurement is also discussed. The measurement results show that this 60GHz mmWave module can provide an acceptable performance only in some cases, which restricts its application scenarios.

In this paper, impact of antenna directivity and bandwidth on the small scale fading statistics have been analyzed for millimeter-wave (mmWave) radio channels. For this purpose, small-scale fading measurements at the mmWave frequency band (58-62 GHz) are carried out using transmit and receive antennas with different antenna directivities (emulated beamforming gains). Measurements emulate a non line-of-sight scenario when the communication between transmit and receive antennas is possible only through a single multipath cluster. In order to compare results, measurements in a line-of-sight scenario with omni-directional antennas are also carried out for reference purpose. Considering two main mmWave system features i.e., high antenna directivity and higher system bandwidth, we report the following results: 1) Randomness/fading in the received signal magnitude vanishes with an increase in bandwidth. 2) The channel impulse response h (t, τ) does not remain a wide-sense stationary (WSS) random process in the slow-time domain i.e., along t, where, the fast-time domain refers to the dimension along τ. 3) Measured channels are WSS in the frequency domain and the coherence bandwidth increases when propagation channels are illuminated with high gain antennas.

Vehicle-to-vehicle (V2V) channels exhibit unique properties due to the highly dynamic environment and low elevation of the antennas at both ends of the link. Of particular importance for the behavior of V2V channels, and consequent reliability of the communication link, is the severity and dynamics of blockage of both the line-of-sight and other multipath components (MPCs). The characteristics of blockage become more important as the carrier frequency increases, and the ability of the signal to penetrate through objects diminishes. To characterize the effects of vehicle blockage, we performed V2V channel measurements in four different frequency bands (6.75, 30, 60, and 73 GHz) in urban and highway scenarios. We analyzed the impact of the blocker size and position on the received power and fast fading parameters, as well as the frequency dependence of these parameters under blockage. Our results show that there is a strong influence of the size of the blocking vehicle on the blockage loss and the angular/delay spread. The position of the blocker relative to the transmitter and receiver also plays an important role. On the other hand, the frequency dependence is quite limited, with the blockage loss increasing slightly and the number of scattered MPCs reducing slightly as frequency increases. The main conclusion of this paper is that V2V communication will be possible in high (millimeter-wave) frequencies, even in the case of blockage by other vehicles.

In the present paper we introduce unique double-directional dual-polarized measurements at 190 GHz in a conference room with the aim of characterizing propagation for channel modelling and beam-forming applications. Assisted by ray-tracing, multiple scatterers have been identified, showing a rich multi-path environment. Investigations have shown that polarization diversity increases spatial diversity and a more deterministic modelling approach in polarization is needed to avoid overestimating polarization diversity gains.

Industry 4.0 is the scenario in which the 5G and beyond networks are expected to show all their potential. However, while propagation at sub-6 GHz has been widely investigated in industry environments, mm-waves propagation and channel modelling in those scenarios is still under early research. Therefore, we introduce novel simultaneous multiband ultra-wideband measurements at 6.75 GHz and 30 GHz in LOS and NLOS with RX below and above clutter level. This unique set-up allows a direct comparison between the sub-6 GHz and mm-wave channel. Results have shown larger specular to dense multi-path components power ratio and shorter large-scale parameters at mm-waves.

This contribution presents propagation measurements at 60 GHz in indoor non-line of sight scenarios. Besides analysing the measurements, the authors have considered several system aspects, such as bandwidth and directivity in order to investigate the small-scale behaviour of the radio channel after beam-forming. The results show that due to the high resolution in the angular and time domain achievable by large antenna arrays and wideband signals, the fading is significantly reduced and the equivalent transmission channel becomes more sparse and deterministic. These insights represent an essential source of information for future systems design and the ongoing channel modelling activities for systems operating at mm-waves.

This paper presents 70–77 GHz wideband channel characteristics of delay, spatial, and polarimetric domains for small office and entrance hall scenarios. Based on measured multidimensional power spectra of delay, and directions (i.e., azimuth and elevation) of departure and of arrival, multipath clusters were detected using the K-means, threshold-based, and Gaussian-mixture-model clustering algorithms. The spatial positions and directions of clusters, including the first- and last-hop scatterers, are determined using a measurement-based ray tracer method. Our experimental analysis reveals that the composite and cluster-level channel behaviors are considerably dependent on the geometry of environments, the interaction/bouncing order between a wave and objects, the position of transmitter and receiver, the underlying antenna type, and the polarimetric combination. The angular spread of arrival is observed larger than that of departure, and the characteristics for the horizontal polarization are comparable with those for vertical polarization. Our results also manifest the merits of threshold-based clustering algorithm in terms of clusters’ compactness, separation, and exclusiveness. We further find that the identified interaction points exhibit excellent agreement with real physical objects existing in the environment. Based on these results, an indoor stochastic channel model is established not only for the composite parameters and cluster characteristics, but also for the first- and last-hop scatterers’ position, direction, and their dispersive statistics. This model can be used to generate channel realizations of reasonable spatial consistency.

Myriad factors influence the initial phase of treatment. Among these, the impact of the referral source and what transpires during the referral process is a neglected area. The patient, therapist, and referral source will inevitably introduce their unique selves into dynamic interactions. This article focuses on the experiences of the therapist. Several factors, including economic and personal ones, contribute to how vulnerable a therapist will be to the ubiquitous pull of self-interest. In addition, the thinking and behavior of each participant—patient, therapist, referral source—may be further shaped by the contextual fields within which these individuals exist and in which treatment takes place. The term “referral matrix” is proposed to capture the dynamic interplay of the constellation of factors within and surrounding the three participants, which may manifest at conscious and unconscious levels. Case studies are presented to demonstrate how the various influences can intertwine. The author d...

Humans and robots are increasingly sharing theirworkspaces to benefit from the precision, endurance, andstrength of machines and the universal capabilities of humans.Instead of performing time-consuming real experiments, com-puter simulations of humans could help to optimally orchestratehuman and robotic tasks—either for setting up new productioncells or by optimizing the motion planning of already installedrobots. Especially when human-robot coexistence is optimizedusing machine learning, being able to synthesize a huge numberof human motions is indispensable. However, no solution existsthat automatically creates a range of human motions from ahigh-level specification of tasks. We propose a novel method thatautomatically generates human motions from linear temporallogic specifications and demonstrate our approach by numericalexamples.

Development of a comprehensive channel propagation model for high-fidelity design and deployment of wireless communication networks necessitates an exhaustive measurement campaign in a variety of operating environments and with different configuration settings. As the campaign is time-consuming and expensive, the effort is typically shared by multiple organizations, inevitably with their own channel-sounder architectures and processing methods. Without proper benchmarking, it cannot be discerned whether observed differences in the measurements are actually due to the varying environments or to discrepancies between the channel sounders themselves. The simplest approach for benchmarking is to transport participant channel sounders to a common environment, collect data, and compare results. Because this is rarely feasible, this paper proposes an alternative methodology – which is both practical and reliable – based on a mathematical system model to represent the channel sounder. The model parameters correspond to the hardware features specific to each system, characterized through precision, in situ calibration to ensure accurate representation; to ensure fair comparison, the model is applied to a ground-truth channel response that is identical for all systems. Five worldwide organizations participated in the cross-validation of their systems through the proposed methodology. Channel sounder descriptions, calibration procedures, and processing methods are provided for each organization as well as results and comparisons for 20 ground-truth channel responses.

The Protvino accelerator facility located in the Moscow region, Russia, is in a good position to offer a rich experimental research program in the field of neutrino physics. Of particular interest is the possibility to direct a neutrino beam from Protvino towards the KM3NeT/ORCA detector, which is currently under construction in the Mediterranean Sea 40 km offshore Toulon, France. This proposal is known as P2O. Thanks to its baseline of 2595 km, this experiment would yield an unparalleled sensitivity to matter effects in the Earth, allowing for the determination of the neutrino mass ordering with a high level of certainty after only a few years of running at a modest beam intensity of $\approx$ 90 kW. With a prolonged exposure ($\approx$ 1500 kW*yr), a 2$\sigma$ sensitivity to the leptonic CP-violating Dirac phase can be achieved. A second stage of the experiment, comprising a further intensity upgrade of the accelerator complex and a densified version of the ORCA detector (Super-ORCA), would allow for up to a 6$\sigma$ sensitivity to CP violation and a 10$^\circ$-17$^\circ$ resolution on the CP phase after 10 years of running with a 450 kW beam, competitive with other planned experiments. The initial composition and energy spectrum of the neutrino beam would need to be monitored by a near detector, to be constructed several hundred meters downstream from the proton beam target. The same neutrino beam and near detector set-up would also allow for neutrino-nucleus cross section measurements to be performed. A short-baseline sterile neutrino search experiment would also be possible., Comment: 27 pages, 15 figures, 2 tables

We introduce simultaneous multi-band ultra wideband double-directional measurements at 6.75 GHz, 30 GHz, and 60 GHz in a corner scenario at a "T" intersection in an urban environment. We analyse the influence of the dynamics from LOS to NLOS on the large-scale parameters, showing that the propagation channel offers similar scattering opportunities at the sub-6 GHz and the mm-wave bands. In addition, we discuss the possibility of utilizing the directional information from sub-6 GHz to steer beams in the mm-waves bands.