
Deep neural networks have shown promise in collaborative filtering (CF).
However, existing neural approaches are either userbased or itembased, which
cannot leverage all the underlying information explicitly. We propose CFUIcA,
a neural coautoregressive model for CF tasks, which exploits the structural
correlation in the domains of both users and items. The coautoregression
allows extra desired properties to be incorporated for different tasks.
Furthermore, we develop an efficient stochastic learning algorithm to handle
large scale datasets. We evaluate CFUIcA on two popular benchmarks: MovieLens
1M and Netflix, and achieve stateoftheart performance in both rating
prediction and topN recommendation tasks, which demonstrates the effectiveness
of CFUIcA.

Network representation learning (NRL) aims to learn lowdimensional vectors
for vertices in a network. Most existing NRL methods focus on learning
representations from local context of vertices (such as their neighbors).
Nevertheless, vertices in many complex networks also exhibit significant global
patterns widely known as communities. It's intuitive that vertices in the same
community tend to connect densely and share common attributes. These patterns
are expected to improve NRL and benefit relevant evaluation tasks, such as link
prediction and vertex classification. Inspired by the analogy between network
representation learning and text modeling, we propose a unified NRL framework
by introducing community information of vertices, named as Communityenhanced
Network Representation Learning (CNRL). CNRL simultaneously detects community
distribution of each vertex and learns embeddings of both vertices and
communities. Moreover, the proposed community enhancement mechanism can be
applied to various existing NRL models. In experiments, we evaluate our model
on vertex classification, link prediction, and community detection using
several realworld datasets. The results demonstrate that CNRL significantly
and consistently outperforms other stateoftheart methods while verifying our
assumptions on the correlations between vertices and communities.

Recent work has found that the metalpoor star HD 94028 shows interesting
abundance features. The sprocess material (e.g., Ba, La, Ce, and Pb) and
rprocess material (e.g., Eu, Os, Pt) are mildly overabundant while the element
C is not enhanced. However, the observed supersolar ratio [As/Ge](= $0.99 \pm
0.23$) and subsolar ratio [Se/As](= $0.16 \pm 0.22$) could not be fitted by
the combination of s and rprocess abundances. In this work, adopting the
abundance decomposition method, we investigate the astrophysical origins of the
elements in this star. We find that the primary process of massive stars
dominates the production of the light elements and irongroup elements. The
lighter neutroncapture elements from Ge to Te mainly come from the weak
rprocess. The elements from Ba to Pb are synthesized dominantly by the main s
and main rprocess. The observed abundance features of Ge, As, and Se in HD
94028 are mainly produced by the weak rprocess, which possesses the features
of the supersolar ratio [As/Ge] and subsolar ratio [Se/As]. Because Ge is not
synthesized solely by the neutroncapture process, As should be the lightest
neutroncapture element. Although the iprocess has been suggested in several
astrophysical environments, it should be superfluous to explaining the
abundances of the lighter neutroncapture elements in HD 94028.

The abundance scatter of heavier rprocess elements (Z > 56) relative to Fe
([r/Fe]) in metalpoor stars preserves excellent information of the star
formation history and provides important insights into the various situations
of the Galactic chemical enrichment. In this respect, the upper and lower
boundaries of [r/Fe] could present useful clues for investigating the extreme
situations of the star formation history and the early Galactic chemical
evolution. In this paper, we investigate the formation of the upper and lower
boundaries of [r/Fe] for the gas clouds. We find that, for a cloud from which
metalpoor stars formed, the formation of the upper limits of [r/Fe] is mainly
due to the pollution from a single main rprocess event. For a cloud from which
metalpoor stars formed, the formation of the lower limits of [r/Fe] is mainly
due to the pollution from a single SN II event that ejects primary Fe.

Traditional intelligent fault diagnosis of rolling bearings work well only
under a common assumption that the labeled training data (source domain) and
unlabeled testing data (target domain) are drawn from the same distribution.
However, in many realworld applications, this assumption does not hold,
especially when the working condition varies. In this paper, a new adversarial
adaptive 1D CNN called A2CNN is proposed to address this problem. A2CNN
consists of four parts, namely, a source feature extractor, a target feature
extractor, a label classifier and a domain discriminator. The layers between
the source and target feature extractor are partially untied during the
training stage to take both training efficiency and domain adaptation into
consideration. Experiments show that A2CNN has strong faultdiscriminative and
domaininvariant capacity, and therefore can achieve high accuracy under
different working conditions. We also visualize the learned features and the
networks to explore the reasons behind the high performance of our proposed
model.

This paper is concerned with the inverse scattering problem by an unbounded
rough surface. A direct imaging method is proposed to reconstruct the rough
surface from the scattered nearfield Cauchy data generating by point sources
and measured on a horizontal straight line segment at a finite distance above
the rough surface. Theoretical analysis of the imaging algorithm is given for
the case of a penetrable rough surface, but the imaging algorithm also works
for impenetrable surfaces with Dirichlet or impedance boundary conditions.
Numerical experiments are presented to show that the direct imaging algorithm
is fast, accurate and very robust with respect to noise in the data.

A new laboratory bound on the axialvector mediated interaction between
electron spins at micrometer scale is established with single nitrogenvacancy
centers in diamond. A single crystal of pterphenyl doped pentancened$_{14}$
under laser pumping provides the source of polarized electron spins. Based on
the measurement of polarization signal via nitrogenvacancy centers, we set a
constraint for the exotic electronelectron coupling, $g_A^eg_A^e$, within the
force range from 10 to 900 $\mu$m. The obtained upper bound of the coupling at
500 $\mu$m is $g_A^eg_A^e / 4\pi\hbar c \leq 5.7\times 10^{19} $, which is
one order of magnitude more stringent than previous experiment. Our result
shows that the NV center can be a promising platform for searching for new
particles predicted by theories beyond the standard model.

Seismic horizons are geologically significant surfaces that can be used for
building geology structure and stratigraphy models. However, horizon tracking
in 3D seismic data is a timeconsuming and challenging problem. Relief human
from the tedious seismic interpretation is one of the hot research topics. We
proposed a novel automatically seismic horizon tracking method by using a deep
convolutional neural network. We employ a stateofart endtoend semantic
segmentation method to track the seismic horizons automatically. Experiment
result shows that our proposed neural network can automatically track multiple
horizons simultaneously. We validate the effectiveness and robustness of our
proposed method by comparing automatically tracked horizons with manually
picked horizons.

This paper is concerned with the inverse scattering problem of timeharmonic
elastic waves by an unbounded rigid rough surface. A direct imaging method is
developed to reconstruct the unbounded rough surface from the elastic scattered
nearfield Cauchy data generated by point sources. A HelmholtzKirchhofftype
identity is derived and then used to provide a theoretical analysis of the
direct imaging algorithm. Numerical experiments are presented to show that the
direct imaging algorithm is fast, accurate and robust with respect to noise in
the data.

Quantum sensing based on nitrogenvacancy (NV) centers in diamond has been
developed as a powerful tool for microscopic magnetic resonance. However, the
reported sensortosample distance is limited within tens of nanometers because
the signal of spin fluctuation decreases cubically with the increasing
distance. Here we extend the sensing distance to tens of micrometers by
detecting spin polarization rather than spin fluctuation. We detected the
mesoscopic magnetic resonance spectra of polarized electrons of a
pentacenedoped crystal, measured its two typical decay times and observed the
optically enhanced spin polarization. This work paves the way for the NVbased
mesoscopic magnetic resonance spectroscopy and imaging at ambient conditions.

We propose Graphical Generative Adversarial Networks (GraphicalGAN) to model
structured data. GraphicalGAN conjoins the power of Bayesian networks on
compactly representing the dependency structures among random variables and
that of generative adversarial networks on learning expressive dependency
functions. We introduce a structured recognition model to infer the posterior
distribution of latent variables given observations. We propose two alternative
divergence minimization approaches to learn the generative model and
recognition model jointly. The first one treats all variables as a whole, while
the second one utilizes the structural information by checking the individual
local factors defined by the generative model and works better in practice.
Finally, we present two important instances of GraphicalGAN, i.e. Gaussian
Mixture GAN (GMGAN) and State Space GAN (SSGAN), which can successfully learn
the discrete and temporal structures on visual datasets, respectively.

This paper is concerned with the inverse timeharmonic elastic scattering
problem of recovering unbounded rough surfaces in two dimensions. We assume
that elastic plane waves with different directions are incident onto a rigid
rough surface in a half plane. The elastic scattered field is measured on a
horizontal straight line segment within a finite distance above the rough
surface. A direct imaging algorithm is proposed to recover the unbounded rough
surface from the scattered nearfield data, which involves only inner products
between the data. Numerical experiments are presented to show that the
inversion scheme is not only efficient but also accurate and robust with
respect to noise.

We report astrometric results of VLBI phasereferencing observations of 22
GHz \hho\ masers emission toward the red hypergiant \vxsgr, one of most massive
and luminous red hypergiant stars in our Galaxy, using the Very Long Baseline
Array. A background source, \Jtwoze, projected 4\d4 from the target \vxsgr, was
used as the phase reference. For the low declinations of these sources, such a
large separation normally would seriously degrade the relative astrometry. We
use a twostep method of tropospheric delay calibration, which combines the
VLBI geodeticblock (or GPS) calibration with an imageoptimization
calibration, to obtain a trigonometric parallax of $0.64\pm0.04$ mas,
corresponding to a distance of 1.56$^{+0.11}_{0.10}$ kpc. The measured proper
motion of \vxsgr\ is $0.36\pm0.76$ and $2.92\pm0.78$ \masy\ in the eastward
and northward directions. The parallax and proper motion confirms that \vxsgr\
belong to the Sgr OB1 association. Rescaling bolometric luminosities in the
literature to our parallax distance, we find the luminosity of \vxsgr~is $(1.95
\pm 0.62) \times 10^5$ \Lsun, where the uncertainty is dominated by differing
photometry measurements.

Stein variational gradient descent (SVGD) is a remarkable recent Bayesian
inference method, which has stronger approximating ability than traditional
variational inference methods, and is more effective than Monte Carlo methods
with the same particle size. However, we observed that SVGD still manifests
particle degeneracy as the dimension increases: particles tend to collapse on
local modes. We take an initial step towards understanding this phenomenon by
analyzing the repulsive force and find that there exists a negative correlation
between the repulsive force and the dimensionality which should be blamed for
this phenomenon. We also propose Message Passing SVGD (MPSVGD) to solve this
problem. By leveraging the conditional independence structure of probabilistic
graphical models (PGMs), MPSVGD converts the original high dimensional global
inference problem into a set of local ones over the Markov blanket with lower
dimensions. Experimental results show its advantages of exploring structural
information over SVGD and particle efficiency and approximation flexibility
over other inference methods on graphical models.

The recently proposed selfensembling methods have achieved promising results
in deep semisupervised learning, which penalize inconsistent predictions of
unlabeled data under different perturbations. However, they only consider
adding perturbations to each single data point, while ignoring the connections
between data samples. In this paper, we propose a novel method, called Smooth
Neighbors on Teacher Graphs (SNTG). In SNTG, a graph is constructed based on
the predictions of the teacher model, i.e., the implicit selfensemble of
models. Then the graph serves as a similarity measure with respect to which the
representations of "similar" neighboring points are learned to be smooth on the
lowdimensional manifold. We achieve stateoftheart results on
semisupervised learning benchmarks. The error rates are 9.89%, 3.99% for
CIFAR10 with 4000 labels, SVHN with 500 labels, respectively. In particular,
the improvements are significant when the labels are fewer. For the
nonaugmented MNIST with only 20 labels, the error rate is reduced from
previous 4.81% to 1.36%. Our method also shows robustness to noisy labels.

Nowadays, the internet of vehicles (IoV) has been evolved into the stage of
vehicle to everything (V2X). However, the majority of existing work focuses on
the motorvehicles. In contrast, the sharing bicycle system is vastly and
rapidly deployed as a feasible internet of things (IoT) application scene for
the last mile problem (e.g., from station to home/office). Moreover, the
internet access of current V2X is relied on the backhaul to roadside unit
(RSU) connections. In this paper, other than prior work, we propose a versatile
V2X system with a distributed framework and heterogeneous caching method. All
the vehicles and devices ontheroad (motorvehicle, nonmotorvehicle,
pedestrian, etc.) are comprehensively included in the proposed networks. We
further introduce a heterogeneous cache method for effective wireless
transmission while utilizing the massive connected devices. The potential
research trends on achieving high speed transmission, deep programming
dedicated network slicing are highlighted as well as the big data, machine
learning (ML), fog computing based image recognition and reconstruction, to
provide some insights for future studies. Finally, the challenging issues,
i.e., urban street canyon path loss and channel models, ultra reliable
communication and low latency requirements, are discussed.

$^{57}$Fe M\"ossbauer spectra were measured at different temperatures between
5.9 K and 300 K on the recently discovered selfdoped superconducting
RbGd$_2$Fe$_4$As$_4$O$_2$ with T$_c$ as high as 35 K. Singlet pattern was
observed down to the lowest temperature measured in this work, indicating the
absence of static magnetic order on the Fe site. The intermediate isomer shift
in comparison with that of the samples RbFe$_2$As$_2$ and GdFeAsO confirms the
self doping induced local electronic structure change. Surprisingly, we observe
two magnetic fluctuation induced spectral broadenings below $\sim$15 K and
$\sim$100 K which are believed to be originated from the transferred magnetic
fluctuations of the Gd$^{3+}$ moments and that of the magnetic fluctuations of
the Fe atoms, respectively.

We present new criteria for selecting HII regions from the Infrared
Astronomical Satellite (IRAS) Point Source catalogue (PSC), based on an HII
region catalogue derived manually from the allsky Widefield Infrared Survey
Explorer (WISE). The criteria are used to augment the number of HII region
candidates in the Milky Way. The criteria are defined by the linear decision
boundary of two samples: IRAS point sources associated with known HII regions,
which serve as the HII region sample, and IRAS point sources at high Galactic
latitudes, which serve as the nonHII region sample. A machine learning
classifier, specifically a support vector machine (SVM), is used to determine
the decision boundary. We investigate all combinations of four IRAS bands and
suggest that the optimal criterion is log(F$_{\rm 60}$/F$_{\rm
12}$)$\ge$(0.19$\times$log(F$_{\rm 100}$/F$_{\rm 25}$)+ 1.52), with detections
at 60 and 100 micron. This selects 3041 HII region candidates from the IRAS
PSC. We find that IRAS HII region candidates show evidence of evolution on the
twocolour diagram. Merging the WISE HII catalogue with IRAS HII region
candidates, we estimate a lower limit of approximately 10200 for the number of
HII regions in the Milky Way.

We determined the \fodd\ values, $0.46\pm0.08$, $0.51\pm0.09$, $0.50\pm0.13$,
$0.48\pm0.12$, which correspond to the rcontribution 100\% for four rII
stars, \cs, \hen, \hes\, and \het, respectively. Our results suggest that
almost all of the heavy elements (in the range from Ba to Pb) in rII stars
have a common origin, that is, from a single rprocess (the main rprocess). We
found that the \fodd\ has a intrinsic nature, and should keep constant value of
about 0.46 in the main rprocess yields, which is responsible for the heavy
element enhancement of rII stars and of our Galaxy chemical enhancement. In
addition, except the abundance ratio [Ba/Eu] the \fodd\ is also an important
indicator, which can be used to study the relative contributions of the r and
sprocess during the chemical evolution history of the Milky Way and the
enhancement mechanism in stars with peculiar abundance of heavy elements.

The integral equation method is widely used in numerical simulations of 2D/3D
acoustic and electromagnetic scattering problems, that needs a large number of
values of the Green's functions. A significant topic is the scattering problems
in periodic domains, that the corresponding Green's functions are
quasiperiodic. The quasiperiodic Green's functions are defined by series that
converge too slowly to be used for calculations. Many mathematicians have
developed several efficient numerical methods to calculate quasiperiodic
Green's functions. In this paper, we will introduce a new FFTbased fast
algorithm to compute the 2D/3D quasiperiodic Green's functions for both
Helmholtz equations and Maxwell's equations. The convergence results and error
estimates are also investigated in this paper. At the end of this paper, the
numerical examples will be given to show that when large number of values are
needed, the new algorithm is very competitive.

Studies on simulation input uncertainty often built on the availability of
input data. In this paper, we investigate an inverse problem where, given only
the availability of output data, we nonparametrically calibrate the input
models and other related performance measures of interest. We propose an
optimizationbased framework to compute statistically valid bounds on input
quantities. The framework utilizes constraints that connect the statistical
information of the realworld outputs with the inputoutput relation via a
simulable map. We analyze the statistical guarantees of this approach from the
view of datadriven robust optimization, and show how the guarantees relate to
the function complexity of the constraints arising in our framework. We
investigate an iterative procedure based on a stochastic quadratic penalty
method to approximately solve the resulting optimization. We conduct numerical
experiments to demonstrate our performance in bounding the input models and
related quantities.

Directional modulation (DM) is a physical layer security technique based on
antenna arrays and so far the polarisation information has not been considered
in its designs. To increase the channel capacity, we consider exploiting the
polarisation information and send two different signals simultaneously at the
same direction, same frequency, but with different polarisations. These two
signals can also be considered as one composite signal using the four
dimensional (4D) modulation scheme across the two polarisation diversity
channels. In this paper, based on crossdipole arrays, we formulate the design
to find a set of common weight coefficients to achieve directional modulation
for such a composite signal and examples are provided to verify the
effectiveness of the proposed method.

Motivated by the increasing use of online appointment booking platforms, we
study how to offer appointment slots to customers in order to maximize the
total number of slots booked. We develop two models, nonsequential offering
and sequential offering, to capture different types of interactions between
customers and the scheduling system. In these two models, the scheduler offers
either a single set of appointment slots for the arriving customer to choose
from, or multiple sets in sequence, respectively. For the nonsequential model,
we identify a static randomized policy which is asymptotically optimal when the
system demand and capacity increase simultaneously, and we further show that
offering all available slots at all times has a constant factor of 2
performance guarantee. For the sequential model, we derive a closedform
optimal policy for a large class of instances and develop a simple, effective
heuristic for those instances without an explicit optimal policy. By comparing
these two models, our study generates useful operational insights for improving
the current appointment booking processes. In particular, our analysis reveals
an interesting equivalence between the sequential offering model and the
nonsequential offering model with perfect customer preference information.
This equivalence allows us to apply sequential offering in a wide range of
interactive scheduling contexts. Our extensive numerical study shows that
sequential offering can significantly improve the slot fill rate (68% on
average and up to 18% in our testing cases) compared to nonsequential
offering.

We conducted an SiO maser survey towards 221 Orich AGB stars with the aim of
identifying maser emission associated with the Sagittarius stellar stream. In
this survey, maser emission was detected in 44 targets, of which 35 were new
detections. All of these masers are within 5 kpc of the Sun. We also compiled a
Galactic SiO maser catalogue including ~2300 SiO masers from the literature.
The distribution of these SiO masers give a scale height of 0.40 kpc, while 42
sources deviate from the Galactic plane by more than 1.2 kpc, half of which
were found in this survey. Regarding SiO masers in the disc, we found both the
rotational speeds and the velocity dispersions vary with the Galactic plane
distance. Assuming Galactic rotational speed $\Theta$0 = 240 km/s , we derived
the velocity lags are 15 km/s and 55 km/s for disc and offplane SiO masers
respectively. Moreover, we identified three groups with significant peculiar
motions (with 70% confidence). The most significant group is in the thick disc
that might trace stream/peculiar motion of the Perseus arm. The other two
groups are mainly made up of offplane sources. The northern and southern
offplane sources were found to be moving at ~33 km/s and 54 km/s away from the
Galactic plane, respectively. Causes of these peculiar motions are still
unclear. For the two offplane groups, we suspect they are thick disc stars
whose kinematics affected by the Sgr stellar stream or very old Sgr stream
debris.

This paper is concerned with uniqueness in inverse acoustic scattering with
phaseless farfield data at a fixed frequency. The main difficulty of this
problem is the socalled translation invariance property of the modulus of the
farfield pattern generated by one plane wave as the incident field. Based on
our previous work (J. Comput. Phys. 345 (2017), 5873), the translation
invariance property of the phaseless farfield pattern can be broken by using
infinitely many sets of superpositions of two plane waves as the incident
fields at a fixed frequency. In this paper, we prove that the obstacle and the
index of refraction of an inhomogeneous medium can be uniquely determined by
the phaseless farfield patterns generated by infinitely many sets of
superpositions of two plane waves with different directions at a fixed
frequency under the condition that the obstacle is a priori known to be a
soundsoft or nonabsorbing impedance obstacle and the index of refraction $n$
of the inhomogeneous medium is realvalued and satisfies that either $n1\ge
c_1$ or $n1\lec_1$ in the support of $n1$ for some positive constant $c_1$.
To the best of our knowledge, this is the first uniqueness result in inverse
scattering with phaseless farfield data. Our proofs are based essentially on
the limit of the normalized eigenvalues of the farfield operators which is
also established in this paper by using a factorization of the farfield
operators.