Abstract

The tourism industry is important for the benefits it brings and due to its role as a commercial activity that creates demand and growth for many more industries. Yet there is not much work on data science problems in tourism. Unfortunately, there is not even a standard benchmark for evaluation of tourismspecific data science tasks and models. In this paper, we propose a benchmark, TOURISMNLG, of five natural language generation (NLG) tasks for the tourism domain and release corresponding datasets with standard train, validation and test splits1 . Further, previously proposed data science solutions for tourism problems do not leverage the recent benefits of transfer learning. Hence, we also contribute the first rigorously pretrained mT5 and mBART model checkpoints for the tourism domain. The models have been pretrained on four tourism-specific datasets covering different aspects of tourism. Using these models, we present initial baseline results on the benchmark tasks. We hope that the dataset will promote active research for natural language generation for travel and tourism. Keywords: NLG for Tourism · Long QA · Blog-Title Generation · Forum-Title Generation · Paragraph Generation · Short QA

Abstract

Recent approaches to the tensor completion problem have often overlooked the nonnegative structure of the data. We consider the problem of learning a nonnegative lowrank tensor, and using duality theory, we propose a novel factorization of such tensors. The factorization decouples the nonnegative constraints from the low-rank constraints. The resulting problem is an optimization problem on manifolds, and we propose a variant of Riemannian conjugate gradients to solve it. We test the proposed algorithm across various tasks such as colour image inpainting, video completion, and hyperspectral image completion. Experimental results show that the proposed method outperforms many state-of-the-art tensor completion algorithms.

Abstract

Sparse neural networks are shown to give accurate predictions competitive to denser versions, while also minimizing the number of arithmetic operations performed. However current GPU hardware can only exploit structured sparsity patterns for better efficiency. We propose a framework for generating structured multilevel block sparse neural networks by using the theory of graph products. Our Ramanujan bipartite graph product (RBGP) framework uses products of Ramanujan graphs to obtain the best connectivity for a given level of sparsity. This essentially ensures that the i.) the networks has the structured block sparsity for which runtime efficient algorithms exists, ii.) the model gives high prediction accuracy, due to the better expressive power derived from the connectivity of the graph, iii.) the graph data structure has a succinct representation that can be stored efficiently in memory. We use our framework to design a specific connectivity pattern called RBGP4 which makes efficient use of the memory hierarchy available on GPU. We benchmark our approach on image classification and machine translation tasks with an edge (Jetson Nano 2GB) as well as server (V100) GPUs. When compared with commonly used sparsity patterns like unstructured and block, we obtain significant speedups while achieving the same level of accuracy.

Abstract

Digital multimedia tools are becoming increasingly important as the world slowly shifts to an online mode. In this work, we present a low-cost, flexible writing pad that uses a 16×16 pressure sensing matrix based on the piezoresistive thin film of velostat. The writing area is 5 cm × 5 cm with an effective pixel area of 0.06 mm2 . A read out circuit is designed to detect the change in resistance of the velostat pixel using a voltage divider. A microprocessor raster scans through the sensor pixel matrix to obtain a data frame of 256 integers. This data is processed using techniques like squaring and normalising (S&N), Gaussian blurring, and adaptive thresholding to generate a more readable output. The writing pad is able to resolve characters larger than 2 cm in length. The flexible writing pad produces legible output while flexed at bending radius of upto 4 cm. Such flexibility promises to enhance the usability and portability of the writing pad significantly. Index Terms—Flexible PCB, Flexible writing pad, Piezoresistive material, Pressure sensor, Velostat

Abstract

The particle flow Gaussian particle filter (PFGPF) uses an invertible particle flow to generate a proposal density. It approximates the predictive and posterior distributions as Gaussian densities. In this paper, we use a bank of PFGPF filters to construct a Particle flow Gaussian sum particle filter (PFGSPF), which approximates the prediction and posterior as Gaussian mixture model. This approximation is useful in complex estimation problems where a single Gaussian approximation is inadequate. We compare the performance of this proposed filter with the PFGPF and others in challenging numerical simulations. Index Terms— Gaussian mixture, invertible particle flow, Gaussian sum particle filer, Gaussian particle filter, PFGPF.

Abstract

Unde rstan din g th e relationship betwee n phys ica l characte ristics an d acti vit y to kinemati c properties of th e spin e ha s impl ication s fo r sports , occupational tasks, clin ica l diagnosi s an d prognosis. This stud y aime d to ca t egorize spin e kinema tic s from flexion, la teral -bend , an d twis t exercise s by examinin g th e co rrelation s to body -mass inde x (BMI), wais t -to -hi p rati o (WHR), an d phys ica l acti vit y le vel s in health y youn g adults . Th e data wa s co llected from a marker -base d optoelectronic motion ca pture setu p an d a marker -less motion ca pture with th e RG B ca m - er a of a standard mobile phone. Co mpa rin g th e accuracies of th e tw o techniques on th e same data se t is impo r - tant to develo p inexpe nsive (RGB ca mera) diagno stics . Sixt y -tw o pa rti c ipant s (4 0 Male , 22 female ) pa rti c ipate d in th e study. Th e angula r di splac ement of th e spin e fo r each exercise wa s extracte d by ve cto r anal ysi s usin g a si n - gl e re ference node /marker . Th e anal ysi s showed no si gni ficant co rrelation s of angula r di splac ement s va lue s fo r th e thre e movement s with BM I or WHR. Th e phys ica l acti vit y leve l of male pa rti c ipant s is si gni ficantly co rrelate d with angula r di splac ement s fo r th e flexio n (p < 0.001) an d la teral -bend (p -valu e ranges from 0.00 1 to 0.04 ) an d weakly co rrelate d (but no t si gni ficant ) fo r twis t exercises. Th e phys ica l acti vit y leve l show s a si gni ficant co rrela - tion fo r only flexio n (p < 0.001) in female pa rti c ipants. Th e findings emph asize th e crit ica l role of phys ica l acti v - it y on mu sculoskel eta l flex ibi lit y in young, health y adults . Th e angles estimate d usin g both techniques were co m - parabl e an d si gni ficantly co rrelate d across pa rti c ipant s an d exercises.

Abstract

Unde rstan din g th e relationship betwee n phys ica l characte ristics an d acti vit y to kinemati c properties of th e spin e ha s impl ication s fo r sports , occupational tasks, clin ica l diagnosi s an d prognosis. This stud y aime d to ca t egorize spin e kinema tic s from flexion, la teral -bend , an d twis t exercise s by examinin g th e co rrelation s to body -mass inde x (BMI), wais t -to -hi p rati o (WHR), an d phys ica l acti vit y le vel s in health y youn g adults . Th e data wa s co llected from a marker -base d optoelectronic motion ca pture setu p an d a marker -less motion ca pture with th e RG B ca m - er a of a standard mobile phone. Co mpa rin g th e accuracies of th e tw o techniques on th e same data se t is impo r - tant to develo p inexpe nsive (RGB ca mera) diagno stics . Sixt y -tw o pa rti c ipant s (4 0 Male , 22 female ) pa rti c ipate d in th e study. Th e angula r di splac ement of th e spin e fo r each exercise wa s extracte d by ve cto r anal ysi s usin g a si n - gl e re ference node /marker . Th e anal ysi s showed no si gni ficant co rrelation s of angula r di splac ement s va lue s fo r th e thre e movement s with BM I or WHR. Th e phys ica l acti vit y leve l of male pa rti c ipant s is si gni ficantly co rrelate d with angula r di splac ement s fo r th e flexio n (p < 0.001) an d la teral -bend (p -valu e ranges from 0.00 1 to 0.04 ) an d weakly co rrelate d (but no t si gni ficant ) fo r twis t exercises. Th e phys ica l acti vit y leve l show s a si gni ficant co rrela - tion fo r only flexio n (p < 0.001) in female pa rti c ipants. Th e findings emph asize th e crit ica l role of phys ica l acti v - it y on mu sculoskel eta l flex ibi lit y in young, health y adults . Th e angles estimate d usin g both techniques were co m - parabl e an d si gni ficantly co rrelate d across pa rti c ipant s an d exercises.

Abstract

Micropumps are one of the most widely used components in lab-on-a-chip devices and other biomedical applications as they can manipulate fluid flow in a very small footprint. In this study, a flexible nozzle-diffuser planar micropump was fabricated using polydimethylsiloxane (PDMS). The pump was fabricated using a two-layer structure: bottom layer with molded channels and chamber encapsulated by a planar top layer. The use of PDMS as the structural material for both layers imparts flexibility and transparency to the complete pump assembly. The micropump was characterized by counting the number of compression cycles required for a known volume of fluid, here water, to flow from the inlet to the outlet. A similar characterization was performed while bending the pump at bending radii of 71 cm and 45 cm. The pump produced a flow rate per compression cycle of 5.53 µl on flat surface and that of 4µl, 3.6 µl when placed on surfaces of bending radii of 71cm and 45 cm respectively. The transparency of the pump and biocompatability of PDMS make it an excellent material for biomedical and colorimetric applications.

Abstract

GPCRs are the most prominent family of membrane proteins that serve as major targets for one-third of the drugs produced. A detailed understanding of the molecular mechanism of drug-induced activation and inhibition of GPCRs is crucial for the rational design of novel therapeutics. The binding of the neurotransmitter adrenaline to the β2-adrenergic receptor (β2AR) is known to induce a flight or fight cellular response, but much remains to be understood about binding-induced dynamical changes in β2AR and adrenaline. In this article, we examine the potential of mean force (PMF) for the unbinding of adrenaline from the orthosteric binding site of β2AR and the associated dynamics using umbrella sampling and molecular dynamics (MD) simulations. The calculated PMF reveals a global energy minimum, which corresponds to the crystal structure of β2AR-adrenaline complex, and a metastable state in which the adrenaline is moved slightly deeper into the binding pocket with a different orientation compared to that in the crystal structure. The orientational and conformational changes in adrenaline during the transition between these two states and the underlying driving forces of this transition are also explored. Based on clustering of MD configurations and machine learning-based statistical analyses of time series of relevant collective variables, the structures and stabilizing interactions of these two states of the β2AR-adrenaline complex are also investigated.

Abstract

—With Cardiovascular Diseases on the rise around the world, Electrocardiograms (ECGs) play a crucial role in their diagnosis owing to their non-invasive nature and simplicity. Medical professionals typically use 12-lead ECGs for medical analysis but gathering 12-lead ECG data is an arduous task outside clinical setting. Modern wearables can collect an ECG with fewer leads than the standard 12 leads. However, medical professionals and conventional ECG analysis software find this reduced lead set data challenging to interpret. By using the reduced lead set data to create standard 12-lead ECG data, ECG reconstruction can solve this issue. This paper proposes a novel single-lead to multi-lead ECG reconstruction solution using a modified Attention U-net framework. Using only the lead II of ECG, our model is able to reproduce the other 11 leads of conventional 12-lead ECG with a Pearson correlation, Mean square error and R-squared error of 0.805, 0.0122 and 0.639, respectively. Further, a single combined model is used to reconstruct all 11 leads simultaneously, improving performance and simultaneously reducing the computational resources needed for training compared to current literature in the field. In comparison to previous works, which only reconstruct small ECG segments, our model is trained to reconstruct longer 10-second ECG signals. We demonstrate our model’s ability for real-life utilisation using a cardiovascular disease classification task. A deep learning model was trained for multi-disease classification on actual 12-lead ECG data and was tested on both original and reconstructed 12-lead ECG signals. The classification accuracies for the original and reconstructed signals were comparable, portraying that our reconstruction model can preserve diagnostically relevant artefacts in its reconstructed signals. This work provides a new promising solution in the field of single-lead ECG reconstruction, taking us a step closer to bridging the divide between reduced lead set data and existing 12-lead ECG end users like clinicians and automatic ECG classifiers. Index Terms—Healthcare, Cardiology, Electrocardiograms, Machine Learning, Neural Networks