Face recognition techniques have been developed significantly in recent years. However, recognizing faces with partial occlusion is still challenging for existing face recognizers, which is heavily ...desired in real-world applications concerning surveillance and security. Although much research effort has been devoted to developing face de-occlusion methods, most of them can only work well under constrained conditions, such as all of faces are from a pre-defined closed set of subjects. In this paper, we propose a robust LSTM-Autoencoders (RLA) model to effectively restore partially occluded faces even in the wild. The RLA model consists of two LSTM components, which aims at occlusion-robust face encoding and recurrent occlusion removal respectively. The first one, named multi-scale spatial LSTM encoder, reads facial patches of various scales sequentially to output a latent representation, and occlusion-robustness is achieved owing to the fact that the influence of occlusion is only upon some of the patches. Receiving the representation learned by the encoder, the LSTM decoder with a dual channel architecture reconstructs the overall face and detects occlusion simultaneously, and by feat of LSTM, the decoder breaks down the task of face de-occlusion into restoring the occluded part step by step. Moreover, to minimize identify information loss and guarantee face recognition accuracy over recovered faces, we introduce an identity-preserving adversarial training scheme to further improve RLA. Extensive experiments on both synthetic and real data sets of faces with occlusion clearly demonstrate the effectiveness of our proposed RLA in removing different types of facial occlusion at various locations. The proposed method also provides significantly larger performance gain than other de-occlusion methods in promoting recognition performance over partially-occluded faces.
Intelligent reflecting surface (IRS) has emerged as a promising paradigm to improve the capacity and reliability of a wireless communication system by smartly reconfiguring the wireless propagation ...environment. To achieve the promising gains of IRS, the acquisition of the channel state information (CSI) is essential, which however is practically difficult since the IRS does not employ any transmit/receive radio frequency (RF) chains in general and it has limited signal processing capability. In this paper, we study the uplink channel estimation problem for an IRS-aided multiuser single-input multi-output (SIMO) system. The existing channel estimation approach for IRS-aided multiuser systems mainly consists of three phases, where the direct channels from the base station (BS) to all the users, the reflected channel from the BS to a typical user via the IRS, and the other reflected channels are estimated sequentially based on the estimation results of the previous phases. However, this approach will lead to a serious error propagation issue, i.e., the channel estimation errors in the first and second phases will deteriorate the estimation performance in the second and third phases. To resolve this difficulty, we propose a novel two-phase channel estimation (2PCE) strategy which is able to alleviate the negative effects caused by error propagation and enhance the channel estimation performance with the same amount of channel training overhead as in the existing approach. Specifically, in the first phase, the direct and reflected channels associated with a typical user are estimated simultaneously by varying the reflection patterns at the IRS, such that the estimation errors of the direct channel associated with this typical user will not affect the estimation of the corresponding reflected channel. In the second phase, we estimate the CSI associated with the other users and demonstrate that by properly designing the pilot symbols of the users and the reflection patterns at the IRS, the direct and reflected channels associated with each user can also be estimated simultaneously, which helps to reduce the error propagation. Moreover, the asymptotic mean squared error (MSE) of the proposed 2PCE strategy is analyzed when the least-square (LS) channel estimation method is employed, and we show that the 2PCE strategy can outperform the existing approach. Finally, extensive simulation results are presented to validate the effectiveness of our proposed channel estimation strategy.
Iridium(III) complexes are an important group of photosensitizers for photodynamic therapy (PDT). This work constructs a donor–acceptor–donor structure‐based iridium(III) complex (IrDAD) with high ...reactive oxygen species (ROS) generation efficiency, negligible dark toxicity, and synergistic PDT and photothermal therapy (PTT) effect under near‐infrared (NIR) stimulation. This complex self‐assembles into metallosupramolecular aggregates with a unique aggregation‐induced PDT behavior. Compared with conventional iridium(III) photosensitizers, IrDAD not only achieves NIR light deep tissue penetration but also shows highly efficient ROS and heat generation with ROS quantum yield of 14.6% and photothermal conversion efficiency of 27.5%. After conjugation with polyethylene glycol (PEG), IrDAD is formulated to a nanoparticulate system (IrDAD‐NPs) with good solubility. In cancer phototherapy, IrDAD‐NPs preferentially accumulate in tumor area and display a significant tumor inhibition in vivo, with 96% reduction in tumor volume, and even tumor elimination.
A donor‐acceptor‐donor structure‐based iridium(III) complex is synthesized for synergistic photodynamic and photothermal therapy of cancer. The complex can be triggered with 808 nm light, generate O2−• to relieve the oxygen‐dependence, and exbibit efficient reactive oxygen species (ROS) and heat generation with a ROS quantum yield of 14.6% and photothermal conversion efficiency of 27.5%.
Microplastics (MPs) have caused increasing global concerns due to their detrimental effects on marine ecosystems. However, the role of photodegradation in altering toxicity of MPs to marine organisms ...is poorly understood. We therefore investigated the photolytic transformation of pristine polystyrene fragments (P-PS) by 60-day ultraviolet (UV) irradiation, and compared the toxicity of P-PS, photodegraded PS (PD-PS), and commercially available polystyrene microbeads (C-PS) to juvenile grouper (Epinephelus moara). Photodegradation reduced the size from ∼55.9 μm of P-PS to ∼38.6 μm of PD-PS, even produced nanoparticles (∼75 nm) with a yield of 7.03 ± 0.37% (w/w), and induced surface oxidation and formation of persistent free radicals (e.g., CO•, COO•). Also, endogenous pollutants (chemical additives and polymer fragments) were leached out. Thus, PD-PS had the highest growth inhibition and lipidosis-driven hepatic lesions of grouper, followed by P-PS and C-PS, which was mainly explained by increased hepatic bioaccumulation of MPs/NPs and released endogenous toxicants. Furthermore, oxidative stress-triggered mitochondrial depolarization, suppression of fatty acid oxidation and transport, and promotion of inflammation were identified as the key mechanisms for the enhanced hepatotoxicity after photodegradation. This work provides new insight into the potential hazard and harm of MPs in marine environments after photodegradation.
In this work, we consider the use of model-driven deep learning techniques for massive multiple-input multiple-output (MIMO) detection. Compared with conventional MIMO systems, massive MIMO promises ...improved spectral efficiency, coverage and range. Unfortunately, these benefits are at the expense of significantly increased computational complexity. To reduce the complexity of signal detection and guarantee the performance, we present a learned conjugate gradient descent network (LcgNet), which is constructed by unfolding the iterative conjugate gradient descent (CG) detector. In the proposed network, instead of calculating the exact values of the scalar step-sizes, we explicitly learn their universal values. Also, we can enhance the proposed network by augmenting the dimensions of these step-sizes. Furthermore, in order to reduce the memory costs, a novel quantized LcgNet is proposed, where a low-resolution nonuniform quantizer is used to quantize the learned parameters. The quantizer is based on a specially designed soft staircase function with learnable parameters to adjust its shape. Meanwhile, due to fact that the number of learnable parameters is limited, the proposed networks are relatively easy to train. Numerical results demonstrate that the proposed network can achieve promising performance with much lower complexity.
Herein, we report the design and synthesis of a mitochondria‐specific, 808 nm NIR light‐activated photodynamic therapy (PDT) system based on the combination of metal–organic frameworks (MOFs) and ...upconversion photochemistry with an organelle‐targeting strategy. The system was synthesized through the growth of a porphyrinic MOF on Nd3+‐sensitized upconversion nanoparticles to achieve Janus nanostructures with further asymmetric functionalization of the surface of the MOF domain. The PDT nanoplatform allows for photosensitizing with 808 nm NIR light, which could effectively avoid the laser‐irradiation‐induced overheating effect. Furthermore, mitochondria‐targeting could amplify PDT efficacy through the depolarization of the mitochondrial membrane and the initiation of intrinsic apoptotic pathway. This work sheds light on the hybrid engineering of MOFs to combat their current limitations for PDT.
Two birds with one stone: Janus upconversion metal–organic framework nanostructures have been developed that allow both mitochondria‐specific delivery and 808 nm NIR‐light‐triggered singlet oxygen generation, leading to amplified photodynamic therapy with minimized overheating effect from the laser.
Nerve injury induces changes in gene transcription in dorsal root ganglion (DRG) neurons, which may contribute to nerve injury-induced neuropathic pain. DNA methylation represses gene expression. ...Here, we report that peripheral nerve injury increases expression of the DNA methyltransferase DNMT3a in the injured DRG neurons via the activation of the transcription factor octamer transcription factor 1. Blocking this increase prevents nerve injury-induced methylation of the voltage-dependent potassium (Kv) channel subunit Kcna2 promoter region and rescues Kcna2 expression in the injured DRG and attenuates neuropathic pain. Conversely, in the absence of nerve injury, mimicking this increase reduces the Kcna2 promoter activity, diminishes Kcna2 expression, decreases Kv current, increases excitability in DRG neurons and leads to spinal cord central sensitization and neuropathic pain symptoms. These findings suggest that DNMT3a may contribute to neuropathic pain by repressing Kcna2 expression in the DRG.
In addition to their use in relieving the symptoms of various diseases, ketogenic diets (KDs) have also been adopted by healthy individuals to prevent being overweight. Herein, we reported that ...prolonged KD exposure induced cardiac fibrosis. In rats, KD or frequent deep fasting decreased mitochondrial biogenesis, reduced cell respiration, and increased cardiomyocyte apoptosis and cardiac fibrosis. Mechanistically, increased levels of the ketone body β-hydroxybutyrate (β-OHB), an HDAC2 inhibitor, promoted histone acetylation of the Sirt7 promoter and activated Sirt7 transcription. This in turn inhibited the transcription of mitochondrial ribosome-encoding genes and mitochondrial biogenesis, leading to cardiomyocyte apoptosis and cardiac fibrosis. Exogenous β-OHB administration mimicked the effects of a KD in rats. Notably, increased β-OHB levels and SIRT7 expression, decreased mitochondrial biogenesis, and increased cardiac fibrosis were detected in human atrial fibrillation heart tissues. Our results highlighted the unknown detrimental effects of KDs and provided insights into strategies for preventing cardiac fibrosis in patients for whom KDs are medically necessary.
Tropone or tropolone and its derivatives (here together called troponoids) belong to a family of natural products with a sevenmembered aromatic ring and various side groups. They are mainly ...synthesized by plants and fungi, and most troponoids play roles of antibacterial defenses in these organisms. With an increasingly severe situation of antibiotic resistant bacteria, as well as a requirement for antifungal medicines, troponoids have attracted extensive studies since they have powerful antibacterial and antifungal activity, particularly against antibiotic-resistant bacteria. In addition, many other biological activities such as antiviral, antitumor, antioxidant, antiinflammatory, insecticidal, or enzyme inhibitor activities are associated with troponoids. After extensive studies in the 1960s-70s, interests in natural troponoids dedclined. However, chemical and biomedical studies on troponoids bloom again from the 1990s. To date great progress has been made with troponoid study in terms of identification of new natural troponoids, chemical synthesis and properties, biological activity, biosynthesis and metabolism. Particularly, bioassay-guided screening strategy and structure-activity relation-directed structure modification and drug design has resulted in the synthesis and discovery of many new derivatives. Many of them have great promise to be developed into new medicines for their potent and specific activities. This review presents the recent advances in troponoid studies and highlights multiple faceted biological activities of troponoids, as well as their relationships with chemical structures. Chemistry, biosynthesis, and production via biotechnology of troponoids are also briefly reviewed. Applications of troponoids in daily life, agriculture, medicine, and industry, and the related patents have been considered to further extend our understanding of the increasing impact of troponoids on humans.
This article describes the design and synthesis of MgO‐modified Ni/CaO catalysts for sorption‐enhanced steam reforming of ethanol. The results show that the introduction of MgO effectively increases ...the dispersion of CaO via forming MgCa(CO3)2 precursor. In the prepared MgO‐modified Ni/CaO catalysts, metallic Ni exists around MgO supported on CaO. Both 100% ethanol conversion and >96% hydrogen purity can be stabilized in 10 cycles over the catalyst containing 20 wt% MgO. The interaction between metallic Ni and MgO enhances the sintering resistance of the catalyst. More importantly, reaction pathway studies have confirmed that the formation of CaCO3 hinders the activation of H2O on the Ni/CaO catalyst surface, and thus inhibits the conversion of the reaction intermediates including HCO* and CH
x*. MgO can dissociate H2O to form hydroxyl groups which participate in the conversion of the reaction intermediates, thereby the MgO‐modified Ni/CaO catalysts have better catalytic performance and carbon deposition resistance.