Domain adaptation seeks to mitigate the shift between training on the emph{source} domain and testing on the emph{target} domain. Most adaptation methods rely on the source data by joint optimization over source data and target data. Source-free meth
ods replace the source data with a source model by fine-tuning it on target. Either way, the majority of the parameter updates for the model representation and the classifier are derived from the source, and not the target. However, target accuracy is the goal, and so we argue for optimizing as much as possible on the target data. We show significant improvement by on-target adaptation, which learns the representation purely from target data while taking only the source predictions for supervision. In the long-tailed classification setting, we show further improvement by on-target class distribution learning, which learns the (im)balance of classes from target data.
We present the results of the analysis of five observations of the globular clutser 47 Tucanae (47 Tuc) with eROSITA (extended Roentgen Survey with an Imaging Telescope Array) on board Spektrum-Roentgen-Gamma (Spektr-RG, SRG). The aim of the work is
the study of the X-ray population in the field of one of the most massive globular clusters in our Milky Way. We focused on the classification of point-like sources in the field of 47 Tuc. The unresolved dense core of 47~Tuc (1.7 radius) and also the sources, which show extended emission are excluded in this study. We applied different methods of X-ray spectral and timing analysis together with multi wavelength studies for the classification of the X-rays sources in the field of 47 Tuc. We detected 888 point-like sources in the energy range of 0.2-5.0 keV. We identified 92 background AGNs and 26 foreground stars. One of the foreground stars is classified as a variable M~dwarf. We also classified 23 X-ray sources as members of 47 Tuc, including 13 symbiotic stars, 3 quiescent low mass X-ray binaries, one millisecond pulsar candidate, and one cataclysmic variable. There are also 4 X-ray sources, which can be either a cataclysmic variable or a contact binary. Moreover, we calculated the X-ray luminosity function of 47 Tuc X-ray sources within a radius of 18.8. It shows that the main population of X-ray sources in 47 Tuc has a luminosity <10$^{32}$erg s$^{-1}$ in the energy range of 0.5-2.0 keV. These sources can mainly be candidates for quiescent low mass X-ray binaries and different types of accreting white dwarfs, especially symbiotic stars.
Context: The eROSITA X-ray telescope onboard the Spectrum-Roentgen-Gamma (SRG) satellite has started to observe new X-ray sources over the full sky at an unprecedented rate. Understanding the selection function of the source detection is important to
the subsequent scientific analysis of the eROSITA catalogs. Aims: Through simulations, we test and optimize the eROSITA source detection procedures, and characterize the detected catalog quantitatively. Methods: Taking the eROSITA Final Equatorial-Depth Survey (eFEDS) as an example, we run extensive photon event simulations using our best knowledge of the instrument characteristics, the background spectrum, and the population of astronomical X-ray sources. We analyze the source detection results based on the origin of each photon. Results. The source detection procedure is optimized according to the source detection efficiency. We choose a two-pronged strategy to build the eFEDS X-ray catalogs, creating a main catalog using only the most sensitive band (0.2-2.3 keV) and an independent hard-band selected catalog using multi-band detection in a range up to 5 keV. From the mock catalogs (available with this paper), we measure the catalog completeness and purity, which can be used in both choosing the sample selection thresholds and in further studies of AGN and cluster demography.
We investigate the physical properties--such as the stellar mass, SFR, IR luminosity, X-ray luminosity, and hydrogen column density--of MIR galaxies and AGN at $z < 4$ in the 140 deg$^2$ field observed by SRG/eROSITA through the eFEDS survey. By cros
s-matching the WISE 22 $mu$m (W4)-detected sample and the eFEDS X-ray point-source catalog, we find that 692 extragalactic objects are detected by eROSITA. We have compiled a multiwavelength dataset. We have also performed (i) an X-ray spectral analysis, (ii) SED fitting using X-CIGALE, (iii) 2D image-decomposition analysis using Subaru HSC images, and (iv) optical spectral fitting with QSFit to investigate the AGN and host-galaxy properties. For 7,088 WISE W4 objects that are undetected by eROSITA, we have performed an X-ray stacking analysis to examine the typical physical properties of these X-ray faint and/or probably obscured objects. We find that (i) 82% of the eFEDS-W4 sources are classified as X-ray AGN with $log,L_{rm X} >$ 42 erg s$^{-1}$; (ii) 67% and 24% of the objects have $log,(L_{rm IR}/L_{odot}) > 12$ and 13, respectively; (iii) the relationship between $L_{rm X}$ and the 6 $mu$m luminosity is consistent with that reported in previous works; and (iv) the relationship between the Eddington ratio and $N_{rm H}$ for the eFEDS-W4 sample and a comparison with a model prediction from a galaxy-merger simulation indicates that approximately 5% of the eFEDS-W4 sources in our sample are likely to be in an AGN-feedback phase, in which strong radiation pressure from the AGN blows out the surrounding material from the nuclear region. Thanks to the wide area coverage of eFEDS, we have been able to constrain the ranges of the physical properties of the WISE W4 sample of AGNs at $z < 4$, providing a benchmark for forthcoming studies on a complete census of MIR galaxies selected from the full-depth eROSITA all-sky survey.
Context: After the successful launch of the Spectrum-Roentgen-Gamma (SRG) mission in July 2019, eROSITA, the soft X-ray instrument aboard SRG, performed scanning observations of a large contiguous field, namely the eROSITA Final Equatorial Depth Surv
ey (eFEDS), ahead of the planned four-year all-sky survey. eFEDS yielded a large sample of X-ray sources with very rich multi-band photometric and spectroscopic coverage. Aims: We present here the eFEDS Active Galactic Nuclei (AGN) catalog and the eROSITA X-ray spectral properties of the eFEDS sources. Methods: Using a Bayesian method, we perform a systematic X-ray spectral analysis for all eFEDS sources. The appropriate model is chosen based on the source classification and the spectral quality, and, in the case of AGN, including the possibility of intrinsic (rest-frame) absorption and/or soft excess emission. Hierarchical Bayesian modeling (HBM) is used to estimate the spectral parameter distribution of the sample. Results: X-ray spectral properties are presented for all eFEDS X-ray sources. There are 21952 candidate AGN, which comprise 79% of the eFEDS sample. Despite a large number of faint sources with low photon counts, our spectral fitting provides meaningful measurements of fluxes, luminosities, and spectral shapes for a majority of the sources. This AGN catalog is dominated by X-ray unobscured sources, with an obscured (logNH>21.5) fraction of 10% derived by HBM. The power-law slope of the catalog can be described by a Gaussian distribution of 1.94+-0.22. Above a photon counts threshold of 500, nine out of 50 AGN have soft excess detected. For the sources with blue UV to optical color (type-I AGN), the X-ray emission is well correlated with the UV emission with the usual anti-correlation between the X-ray to UV spectral slope {alpha}_{OX} and the UV luminosity.
In this study, we investigate the X-ray properties of WISE J090924.01+000211.1 (WISEJ0909+0002), an extremely luminous infrared (IR) galaxy (ELIRG) at $z_{rm spec}$= 1.871 in the eROSITA final equatorial depth survey (eFEDS). WISEJ0909+0002 is a WISE
22 $mu$m source, located in the GAMA-09 field, detected by eROSITA during the performance and verification phase. The corresponding optical spectrum indicates that this object is a type-1 active galactic nucleus (AGN). Observations from eROSITA combined with Chandra and XMM-Newton archival data indicate a very luminous ($L$ (2--10 keV) = ($2.1 pm 0.2) times 10^{45}$ erg s$^{-1}$) unobscured AGN with a power-law photon index of $Gamma$ = 1.73$_{-0.15}^{+0.16}$, and an absorption hydrogen column density of $log,(N_{rm H}/{rm cm}^{-2}) < 21.0$. The IR luminosity was estimated to be $L_{rm IR}$ = (1.79 $pm$ 0.09) $times 10^{14}, L_{odot}$ from spectral energy distribution modeling based on 22 photometric data (X-ray to far-IR) with X-CIGALE, which confirmed that WISEJ0909+0002 is an ELIRG. A remarkably high $L_{rm IR}$ despite very low $N_{rm H}$ would indicate that we are witnessing a short-lived phase in which hydrogen gas along the line of sight is blown outwards, whereas warm and hot dust heated by AGNs still exist. As a consequence of eROSITA all-sky survey, $6.8_{-5.6}^{+16}times 10^2$ such X-ray bright ELIRGs are expected to be discovered in the entire extragalactic sky ($|b| > 10^circ$). This can potentially be the key population to constrain the bright-end of IR luminosity functions.
Correctly perceiving micro-expression is difficult since micro-expression is an involuntary, repressed, and subtle facial expression, and efficiently revealing the subtle movement changes and capturing the significant segments in a micro-expression s
equence is the key to micro-expression recognition (MER). To handle the crucial issue, in this paper, we firstly propose a dynamic segmented sparse imaging module (DSSI) to compute dynamic images as local-global spatiotemporal descriptors under a unique sampling protocol, which reveals the subtle movement changes visually in an efficient way. Secondly, a segmented movement-attending spatiotemporal network (SMA-STN) is proposed to further unveil imperceptible small movement changes, which utilizes a spatiotemporal movement-attending module (STMA) to capture long-distance spatial relation for facial expression and weigh temporal segments. Besides, a deviation enhancement loss (DE-Loss) is embedded in the SMA-STN to enhance the robustness of SMA-STN to subtle movement changes in feature level. Extensive experiments on three widely used benchmarks, i.e., CASME II, SAMM, and SHIC, show that the proposed SMA-STN achieves better MER performance than other state-of-the-art methods, which proves that the proposed method is effective to handle the challenging MER problem.
Real-time applications of energy management strategies (EMSs) in hybrid electric vehicles (HEVs) are the harshest requirements for researchers and engineers. Inspired by the excellent problem-solving capabilities of deep reinforcement learning (DRL),
this paper proposes a real-time EMS via incorporating the DRL method and transfer learning (TL). The related EMSs are derived from and evaluated on the real-world collected driving cycle dataset from Transportation Secure Data Center (TSDC). The concrete DRL algorithm is proximal policy optimization (PPO) belonging to the policy gradient (PG) techniques. For specification, many source driving cycles are utilized for training the parameters of deep network based on PPO. The learned parameters are transformed into the target driving cycles under the TL framework. The EMSs related to the target driving cycles are estimated and compared in different training conditions. Simulation results indicate that the presented transfer DRL-based EMS could effectively reduce time consumption and guarantee control performance.
Knowledge transfer is a promising concept to achieve real-time decision-making for autonomous vehicles. This paper constructs a transfer deep reinforcement learning framework to transform the driving tasks in inter-section environments. The driving m
issions at the un-signalized intersection are cast into a left turn, right turn, and running straight for automated vehicles. The goal of the autonomous ego vehicle (AEV) is to drive through the intersection situation efficiently and safely. This objective promotes the studied vehicle to increase its speed and avoid crashing other vehicles. The decision-making pol-icy learned from one driving task is transferred and evaluated in another driving mission. Simulation results reveal that the decision-making strategies related to similar tasks are transferable. It indicates that the presented control framework could reduce the time consumption and realize online implementation.
The XMM-RM project was designed to provide X-ray coverage of the Sloan Digital Sky Survey Reverberation Mapping (SDSS-RM) field. 41 XMM-Newton exposures, placed surrounding the Chandra AEGIS field, were taken, covering an area of 6.13 deg^2 and reach
ing a nominal exposure depth of ~15 ks. We present an X-ray catalog of 3553 sources detected in these data, using a PSF-fitting algorithm and a sample selection threshold that produces a ~5% fraction of spurious sources. In addition to the PSF-fitting likelihood, we calculate a second source reliability measure based on Poisson theory using source and background counts within an aperture. Using the Poissonian likelihood, we select a sub-sample with a high purity and find that it has similar number count profiles to previous X-ray surveys. The Bayesian method NWAY was employed to identify counterparts of the X-ray sources from the optical Legacy and the IR unWISE catalogs, using a 2-dimensional unWISE magnitude-color prior created from optical/IR counterparts of Chandra X-ray sources. A significant number of the optical/IR counterparts correspond to sources with low detection likelihoods, proving the value of retaining the low-likelihood detections in the catalog. 932 of the XMM-RM sources are covered by SDSS spectroscopic observations. 89% of them are classified as AGN, and 71% of these AGN are in the SDSS-RM quasar catalog. Among the SDSS-RM quasars, 80% are detectable at the depth of the XMM observations.
