What’s Happening at CUSP: November 2019

Youcef Fisli, Abdelouahab Rezaiguia, Salah Guenfoud, and Debra Laefer. “Dynamic response of a multi-span, orthotropic bridge deck under moving truck loading with tandem axles.” Diagnostyka (2019). doi:10.29354/diag/113001.

• A new three-dimensional vehicle with tandem axels is developed to determine dynamic response of bridge deck under load applying truck. The vehicle is modeled by a dynamic system with 9 degrees of freedom to accurately simulate the disposition and the intensity of loads on the bridge deck. The bridge deck is modeled by an orthotropic multi-span plate. The road surface irregularities are modeled by a random function characterized by a spectral roughness coefficient and power spectral density. The modal method is used to solve the equation of motion of the bridge deck. Equations of motion of the vehicle are obtained using the virtual work principle. The coupled equations of motion vehicle/bridge deck are integrated numerically by Newmark’s method. A computational algorithm is then elaborated to solve the integrated equations of motion with iterative process. A numerical example is presented. The resulting distribution of the Dynamic Amplification Factor (DAF) on the bridge deck does not reflect any particular trend, because high values can be obtained at points where the vertical displacement is small. The DAF is significant only under the interaction force. Thus, the road surface roughness was shown to have a significant influence on the dynamic vehicle/bridge deck interaction forces.

Can Cui, Zhen Wang, Pan He, Shanfeng Yuan, Beibei Niu, Ping Kang, Chaogui Kang. “Escaping from pollution: The effect of air quality on inter-city population mobility in China.” Environmental Research Letters, 2019.

• This study utilizes smartphone-based location data and instrumental variable (IV) regression to fill the gap of how air quality affects population mobility. Results indicated that: The increase of air quality index (AQI) by 100 points will cause a 49.60% increase in population outflows, and a rise of 1 µg/m3 of PM2.5 may cause a 0.47% rise in population outflows; Air pollution of former days can drive people to leave their cities three days or a week later by railway or by road; The effect is heterogeneous among workdays, weekends, and holidays. This study implied that air quality management can be critical for urban tourism and environmental competitiveness.

Vincent Lostanlen, Justin Salamon, Andrew Farnsworth, Steve Kelling, and Juan Pablo Bello.  Robust sound event detection in bioacoustic sensor networks. 2019. PLOS ONE 14(10): e0214168.

• In networks of autonomous recording units, the spatiotemporal variability of ambient noise hampers state-of-the-art sound event detectors (SED). In this article, we develop, benchmark, and combine several machine listening techniques to improve the generalizability of SED across heterogeneous acoustic environments. Combining per-channel energy normalization (PCEN) and context adaptation in a convolutional neural network yields state-of-the-art results that are unmatched by artificial data augmentation alone. We release a pre-trained version of our best performing system under the name of BirdVoxDetect, a ready-to-use detector of avian flight calls in field recordings.
• Dataset: https://zenodo.org/record/1205569
• Software: https://github.com/BirdVox/birdvoxdetect

Changhong Wang, Vincent Lostanlen, Emmanouil Benetos, Elaine Chew. Adaptive Time–Frequency Scattering for Periodic Modulation Recognition in Music Signals. International Society on Music Information Retrieval (ISMIR) conference. Delft, the Netherlands, November 2019.

• Vibratos, tremolos, trills, and flutter tonguing cause periodic modulations in the time–frequency. This paper proposes a new machine listening system to detect and classify such playing techniques from limited labeled data. The system is an accelerated, transposition-invariant instance of the wavelet scattering transform. We report state-of-the-art multiclass detection metrics on a new dataset of Chinese bamboo flute recordings (CBF-periDB) alongside explanatory visualizations of scattering coefficients.
• Dataset: https://zenodo.org/record/3250223

Vincent Lostanlen, Kaitlin Palmer, Elly Knight, Christopher Clark, Holger Klinck, Andrew Farsnworth, Tina Wong, Aurora Cramer, and Juan Pablo Bello. Long-distance detection of bioacoustic events with per-channel energy normalization. Detection and Classification of Acoustic Scenes and Events 2019.

• This paper proposes to perform unsupervised detection of bioacoustic events by pooling the magnitudes of spectrogram frames after per-channel energy normalization (PCEN). We prove that PCEN generalizes logarithm-based spectral flux, yet with a tunable time scale for background noise estimation. In comparison with pointwise logarithm, PCEN reduces false alarm rate by 50x in the near field and 5x in the far field, both on avian and marine bioacoustic datasets. Such improvements come at moderate computational cost and require no human intervention, thus heralding a promising future for PCEN in bioacoustics.
• Dataset: https://zenodo.org/record/1208080
• Software: https://github.com/BirdVox/lostanlen2019dcase

Smart Cities Postdoctoral Associate Chenglu Jin has an upcoming paper accepted in the 2020 ACM Asia Conference on Computer and Communications Security (AsiaCCS ’20). “LiS: Lightweight Signature Schemes for Continuous Message Authentication in Cyber-Physical Systems” introduces a highly efficient digital signature scheme that allows CPS devices to sign their messages securely and continuously.

Nicholas S. Caros and Joseph Y.J. Chow, 2020. Effects of violent crime and vehicular crashes on active mode choice decisions in New York City. Travel Behaviour and Society, 18, 37-45.

• Substantial research has explored the effects of different variables on mode choice with the intent of understanding this behavior so that active modes can be encouraged. This study furthers that effort by investigating the impact of perceived danger from crime on the probability of choosing an active mode of transportation among travellers in New York City from 2009 to 2011. This study uses trip data from the New York Metropolitan Transportation Council Regional Household Travel Survey along with historical crime and vehicle collision data to estimate a random utility model. Traveller demographic information, travel cost and incidence of crime and vehicle collisions involving pedestrians or cyclists are used as explanatory variables in a mixed logit model. The finding implies that travellers could be encouraged to cycle by reducing crime levels, or by being provided an alternative route with less crime. Based on the model results it can be determined that travellers are willing to pay $0.66 for a 1000-point reduction in crime severity. An increase in crime of 1% has a much greater impact on bike ridership (2.11% reduction) than on walking (0.06% reduction). Removing crime completely would improve a traveler’s trip satisfaction by as much as $0.26 per trip. Compared to crime, collision rate has a much stronger impact on bike ridership, with an elasticity of −7.56 (3.6 times higher elasticity than crime).

Vincent Lostanlen, Sripathi Sridhar, Brian McFee, Andrew Farnsworth, Juan Pablo Bello. Learning the Helix Topology of Musical Pitch. arXiv preprint arXiv:1910.10246 (2019).

• To explain the consonance of octaves, music psychologists represent pitch as a helix where azimuth and axial coordinate correspond to pitch class and pitch height respectively. This article addresses the problem of discovering this helical structure from unlabeled audio data. We measure Pearson correlations in the constant-Q transform (CQT) domain to build a K-nearest neighbor graph between frequency subbands. Then, we run the Isomap manifold learning algorithm to represent this graph in a three-dimensional space in which straight lines approximate graph geodesics. Experiments on isolated musical notes demonstrate that the resulting manifold resembles a helix which makes a full turn at every octave. A circular shape is also found in English speech, but not in urban noise. We discuss the impact of various design choices on the visualization: instrumentarium, loudness mapping function, and number of neighbors K.