Search

Influence maximization (IM) is formulated as selecting a set of initial users from a social network to maximize the expected number of influenced users. Researchers have made great progress in designing various traditional methods, and their theoretical design and performance gain are close to a limit. In the past few years, learning-based IM methods have emerged to achieve stronger generalization ability to unknown graphs than traditional ones. However, the development of learning-based IM methods is still limited by fundamental obstacles, including 1) the difficulty of effectively solving the objective function; 2) the difficulty of characterizing the diversified underlying diffusion patterns; and 3) the difficulty of adapting the solution under various node-centrality-constrained IM variants. To cope with the above challenges, we design a novel framework DeepIM to generatively characterize the latent representation of seed sets, and we propose to learn the diversified information diffusion pattern in a data-driven and end-to-end manner. Finally, we design a novel objective function to infer optimal seed sets under flexible node-centrality-based budget constraints. Extensive analyses are conducted over both synthetic and real-world datasets to demonstrate the overall performance of DeepIM. The code and data are available at: https://github.com/triplej0079/DeepIM., Comment: In Proceedings of the 40th International Conference on Machine Learning (ICML 2023), Honolulu, Hawaii, USA. PMLR 202, 2023

Cancer-related fatigue (CRF) is one of the most commonly found problems among the cancer survivors which affect all aspects of quality of life (QOL), but it is a treatable problem. CRF is the highly prevalent and debilitating symptom experienced by most of the cancer patients at diagnosis, which increase during therapy and often seen for considerable period of time (months and years) after well-structured cancer diagnosis. Globally, around 50%–90% of the cancer patients experiences the difficulties of cancer-related fatigue, the latter number resemble to the patients subjected to active anticancer chemotherapy and radiotherapy. CRF has been well-defined as per the guiding principle of the National Comprehensive Cancer Network (NCCN) as “a distressing, persistent, subjective sense of tiredness or exhaustion related to cancer or cancer treatment that is not proportional to recent activity and interferes with usual functioning.” Although a lot of evidence and research findings on CRF have been provided within the last two decades, CRF is still underreported, underdiagnosed, and undertreated. Effective management of CRF is important for patients and challenging for clinicians. Moxibustion appeared an efficacious auxiliary therapeutic method for CRF.

Cancer-related fatigue (CRF) is one of the most commonly found problems among the cancer survivors which affect all aspects of quality of life (QOL), but it is a treatable problem. CRF is the highly prevalent and debilitating symptom experienced by most of the cancer patients at diagnosis, which increase during therapy and often seen for considerable period of time (months and years) after well-structured cancer diagnosis. Globally, around 50%–90% of the cancer patients experiences the difficulties of cancer-related fatigue, the latter number resemble to the patients subjected to active anticancer chemotherapy and radiotherapy. CRF has been well-defined as per the guiding principle of the National Comprehensive Cancer Network (NCCN) as “a distressing, persistent, subjective sense of tiredness or exhaustion related to cancer or cancer treatment that is not proportional to recent activity and interferes with usual functioning.” Although a lot of evidence and research findings on CRF have been provided within the last two decades, CRF is still underreported, underdiagnosed, and undertreated. Effective management of CRF is important for patients and challenging for clinicians. Acupuncture and moxibustion appeared an efficacious auxiliary therapeutic method for CRF.

ObjectiveTo study the regulatory role of GLI1/GLI2, a nuclear transcription factor of the Sonic hedgehog (Shh) signaling pathway, in epithelial–mesenchymal transition (EMT) related to hepatic fibrosis in patients with biliary atresia (BA).MethodsThe messenger RNA (mRNA) and protein expression levels of GLI1/GLI2, Snail/Slug, and other Shh- and EMT-related cytokines were tested in the liver tissues of BA patients and animals. Then, GLI1/GLI2 was silenced and overexpressed in mouse intrahepatic bile duct epithelial cells (mIBECs) and BA animals to investigate changes in the mRNA and protein expression of EMT key factors and liver fibrosis indicators. After silencing and overexpression of GLI1/GLI2, immunofluorescence was used to detect the expression of cytokeratin-19 (CK19) and α-smooth muscle actin (α-SMA) in mIBECs, and hematoxylin and eosin (HE) staining and Masson staining were used to observe the degree of liver fibrosis in the BA animals.ResultsCompared with the control, the mRNA and protein expression levels of GLI2, Snail, vimentin, and α-SMA were significantly increased and those of E-cadherin were significantly decreased in liver tissue from BA patients and animals. Overexpression of GLI2 increased the mRNA and protein expression levels of Snail, vimentin, and α-SMA and that of E-cadherin was significantly decreased in mIBECs and BA animals. After GLI2 silencing, the opposite pattern was observed. Immunofluorescence detection showed enhanced expression of the bile duct epithelial cell marker CK19 in mIBECs after GLI2 silencing and enhanced expression of the mesenchymal cell marker α-SMA after GLI2 overexpression. HE and Masson staining suggested that the GLI2-overexpressing group had a significantly higher degree of fibrosis.ConclusionThe Shh signaling pathway plays an important role in fibrogenesis in BA. GLI2 can significantly regulate EMT in mIBECs and livers of BA mice.

Localizing the source of graph diffusion phenomena, such as misinformation propagation, is an important yet extremely challenging task. Existing source localization models typically are heavily dependent on the hand-crafted rules. Unfortunately, a large portion of the graph diffusion process for many applications is still unknown to human beings so it is important to have expressive models for learning such underlying rules automatically. This paper aims to establish a generic framework of invertible graph diffusion models for source localization on graphs, namely Invertible Validity-aware Graph Diffusion (IVGD), to handle major challenges including 1) Difficulty to leverage knowledge in graph diffusion models for modeling their inverse processes in an end-to-end fashion, 2) Difficulty to ensure the validity of the inferred sources, and 3) Efficiency and scalability in source inference. Specifically, first, to inversely infer sources of graph diffusion, we propose a graph residual scenario to make existing graph diffusion models invertible with theoretical guarantees; second, we develop a novel error compensation mechanism that learns to offset the errors of the inferred sources. Finally, to ensure the validity of the inferred sources, a new set of validity-aware layers have been devised to project inferred sources to feasible regions by flexibly encoding constraints with unrolled optimization techniques. A linearization technique is proposed to strengthen the efficiency of our proposed layers. The convergence of the proposed IVGD is proven theoretically. Extensive experiments on nine real-world datasets demonstrate that our proposed IVGD outperforms state-of-the-art comparison methods significantly. We have released our code at https://github.com/xianggebenben/IVGD., Comment: WWW 2022

Contour error compensation of the Computer Numerical Control (CNC) machine tool is a vital technology that can improve machining accuracy and quality. To achieve this goal, the tracking error of a feeding axis, which is a dominant issue incurring the contour error, should be firstly modeled and then a proper compensation strategy should be determined. However, building the precise tracking error prediction model is a challenging task because of the nonlinear issues like backlash and friction involved in the feeding axis; besides, the optimal compensation parameter is also difficult to determine because it is sensitive to the machining tool path. In this paper, a set of novel approaches for contour error prediction and compensation is presented based on the technologies of deep learning and reinforcement learning. By utilizing the internal data of the CNC system, the tracking error of the feeding axis is modeled as a Nonlinear Auto-Regressive Long-Short Term Memory (NAR-LSTM) network, considering all the nonlinear issues of the feeding axis. Given the contour error as calculated based on the predicted tracking error of each feeding axis, a compensation strategy is presented with its parameters identified efficiently by a Time-Series Deep Q-Network (TS-DQN) as designed in our work. To validate the feasibility and advantage of the proposed approaches, extensive experiments are conducted, testifying that, our approaches can predict the tracking error and contour error with very good precision (better than about 99% and 90% respectively), and the contour error compensated based on the predicted results and our compensation strategy is significantly reduced (about 70%~85% reduction) with the machining quality improved drastically (machining error reduced about 50%).

The initial cyanide (CN-) concentration and amount of co-contaminants in GCTs can inhibit bacterial growth and reduce the CN--degrading ability of bacteria. Several microorganisms can biotransform a wide range of organic and inorganic industrial contaminants into nontoxic compounds. However, active enzymatic CN- metabolism processes are mostly constrained by the physical and chemical characteristics of GCTs. High concentrations of toxic metal co-contaminants, such as, Pb, and Cr, and factors, such as pH, temperature, and oxygen concentration create oxidative stress and limit the CN--degrading potential of cyanotrophic strains. The effects of such external and internal factors on the CN--degrading ability of bacteria hinder the selection of suitable microorganisms for CN- biodegradation. Therefore, understanding the effects of the physicochemical properties of GCTs on cyanobacteria strains can help identify suitable microbes and favorable environmental conditions to promote microbial growth and can also help design efficient CN- biodegradation processes. In this review, we present a detailed analysis of the physicochemical properties of GCTs and their effects on microbial CN- degradation.

Zhuang et al. identify SARS-CoV-2–specific T cell epitopes recognized by CD4+ and CD8+ T cells in BALB/c and C57BL/6 mice. SARS-CoV-2–specific T cells are protective and cross-reactive in vivo, and are shaped by type I IFN signaling., Virus-specific T cells play essential roles in protection against multiple virus infections, including SARS-CoV and MERS-CoV. While SARS-CoV-2–specific T cells have been identified in COVID-19 patients, their role in the protection of SARS-CoV-2–infected mice is not established. Here, using mice sensitized for infection with SARS-CoV-2 by transduction with an adenovirus expressing the human receptor (Ad5-hACE2), we identified SARS-CoV-2–specific T cell epitopes recognized by CD4+ and CD8+ T cells in BALB/c and C57BL/6 mice. Virus-specific T cells were polyfunctional and were able to lyse target cells in vivo. Further, type I interferon pathway was proved to be critical for generating optimal antiviral T cell responses after SARS-CoV-2 infection. T cell vaccination alone partially protected SARS-CoV-2–infected mice from severe disease. In addition, the results demonstrated cross-reactive T cell responses between SARS-CoV and SARS-CoV-2, but not MERS-CoV, in mice. Understanding the role of the T cell response will guide immunopathogenesis studies of COVID-19 and vaccine design and validation.

Multi-task learning is a framework that enforces different learning tasks to share their knowledge to improve their generalization performance. It is a hot and active domain that strives to handle several core issues; particularly, which tasks are correlated and similar, and how to share the knowledge among correlated tasks. Existing works usually do not distinguish the polarity and magnitude of feature weights and commonly rely on linear correlation, due to three major technical challenges in: 1) optimizing the models that regularize feature weight polarity, 2) deciding whether to regularize sign or magnitude, 3) identifying which tasks should share their sign and/or magnitude patterns. To address them, this paper proposes a new multi-task learning framework that can regularize feature weight signs across tasks. We innovatively formulate it as a biconvex inequality constrained optimization with slacks and propose a new efficient algorithm for the optimization with theoretical guarantees on generalization performance and convergence. Extensive experiments on multiple datasets demonstrate the proposed methods' effectiveness, efficiency, and reasonableness of the regularized feature weighted patterns., Comment: 17 pages, 4 figures, v1

The Graph Convolutional Network (GCN) has been successfully applied to many graph-based applications. Training a large-scale GCN model, however, is still challenging: Due to the node dependency and layer dependency of the GCN architecture, a huge amount of computational time and memory is required in the training process. In this paper, we propose a parallel and distributed GCN training algorithm based on the Alternating Direction Method of Multipliers (ADMM) to tackle the two challenges simultaneously. We first split GCN layers into independent blocks to achieve layer parallelism. Furthermore, we reduce node dependency by dividing the graph into several dense communities such that each of them can be trained with an agent in parallel. Finally, we provide solutions for all subproblems in the community-based ADMM algorithm. Preliminary results demonstrate that our proposed community-based ADMM training algorithm can lead to more than triple speedup while achieving the best performance compared with state-of-the-art methods., Comment: accepted by NeurIPS 2021 OPT workshop

Alternating minimization methods have recently been proposed as alternatives to the gradient descent for deep neural network optimization. Alternating minimization methods can typically decompose a deep neural network into layerwise subproblems, which can then be optimized in parallel. Despite the significant parallelism, alternating minimization methods are rarely explored in training deep neural networks because of the severe accuracy degradation. In this paper, we analyze the reason and propose to achieve a compelling trade-off between parallelism and accuracy by a reformulation called Tunable Subnetwork Splitting Method (TSSM), which can tune the decomposition granularity of deep neural networks. Two methods gradient splitting Alternating Direction Method of Multipliers (gsADMM) and gradient splitting Alternating Minimization (gsAM) are proposed to solve the TSSM formulation. Experiments on five benchmark datasets show that our proposed TSSM can achieve significant speedup without observable loss of training accuracy. The code has been released at https://github.com/xianggebenben/TSSM., Comment: ICML 2020 Workshop on "Beyond first-order methods in ML systems"

Cemented soil impermeable projects directly or indirectly contact an underground corrosive environment in most cases, which could cause reduced impermeability and malfunction of the cemented soil. Laboratory simulation permeability tests of cemented soil were conducted under different cement ratios in acidic and alkali environments. The variation in the permeability coefficient of the cemented soil was studied under different curing times, pH, and cement contents. Using scanning electron microscopy (SEM) and energy spectrum analysis, the permeability of cemented soil in acidic and alkali environments was studied at the microstructure level. The test results show that an acidic environment has a major effect on the permeability coefficient of cemented soil with a low cement content, and a greater cement content offers increased resistance to acid erosion. In an alkali environment of pH 9-13, the permeability coefficient of the cemented soil gradually decreases as the curing time increases. The research results provide technical parameters for the impermeability and durability design and application of cemented soil in impermeable projects, which has important engineering implications.

Enhancing the reactivity of mine tailings through the efficient and economical technology and using them as active precursors for alkali-activated materials is essential to effectively alleviate environmental issues caused by huge tailings emissions. In this study, an alkaline hydrothermal technology was proposed for increasing the reactivity of gold mine tailings. X-ray diffraction (XRD) analysis and leaching tests indicated that the crystalline structures of tailings were significantly destroyed and amorphous sodium aluminosilicate was identified as the major hydrothermal activation product, which dissolved in water, released large quantities of active Si and Al species, and created an alkaline environment for the slag activation. Subsequently, one-part alkali-activated materials (OPM) were prepared from activated tailings and slag. To improve the hydration properties of these materials, quicklime (CaO) was employed as an additive. OPM sample exhibited the optimum strengths of 8.4, 28.7, 39.5, and 42.5 MPa at 1, 3, 7, 28 days, respectively, when CaO content was controlled at ∼ 6 wt%. pH tests, XRD, thermogravimetric (TG), and scanning electron microscopy (SEM) analyses revealed that the addition of CaO improved the pH of the pore solution, increased the quantity of calcium-sodium aluminosilicate hydrate gel (C(N)-A-S-H, the main hydration reactant), and generated the compact microstructure, thereby enhancing the strength of the OPM samples. However, the strength was dramatically reduced when CaO content exceeded 6 wt% at later stages, mainly because of the severe carbonation of the sample.

This paper puts forward an automatic welding-scar grinding system with an industrial robot and grinding jigs, aiming at solving the issues of low efficiency, labor intensity, poor working environment, and heterogeneity of grinding quality. The designed system is comprised of a control system (Mitsubishi PLC), an industrial robot, a grinding workstation, an end-grinding actuator, and an automatic tool change magazine. The software of the proposed system is featured by an interactive GUI and automatic tool change capability. A physical prototype is developed to demonstrate the effectiveness of the presented grinding system. Grinding test is conducted and the results indicate that robotic grinding can achieve better surface roughness and have gained an increase of 40% in grinding efficiency in comparison with manual one. In the end, existing limitations and future research are discussed.

In all kinds of optical applications, the precision of central optical distance has significant influence on the desired imaging properties according to the optical design requirements. In this paper, we present a measurement method for the determination of the central distance both in assembled systems and single-lens with high precision, high speed and non-invasive characters. The distance between optical surfaces are optically measured by the mean of swept source domain interferometric system. A fiber-optics system with a central wavelength 1310 nm is built, the central distances of all surfaces within coherence length can be quickly measured at the same time. The system pixel resolution is about 0.1μm calibrated by Mitutoyo standard level-zero gauge and calculated by Fast Fourier Transform (FFT) and zero-padding algorithm. System precision double checked by other gauges is less than 0.3μm. The standard gauge-sets and a lens were measured by this system, the result also verified the high precision. The systematic error is less than 0.3μm and the sensitivity is about 22μm experimentally. The high speed swept light source (100kHz used in system) ensures the quick measuring speed. This measurement method has high precision, high speed, non-invasive and high sensitivity characters, and can be applied in related optical system.

The multi-block ADMM has received much attention from optimization researchers due to its excellent scalability. In this paper, the multi-block ADMM is applied to solve two large-scale problems related to isotonic regression. Numerical experiments show that the multi-block ADMM is convergent when the chosen parameter is small enough and the multi-block ADMM scales well compared with baselines., Comment: Accepted by 11th Workshop on Optimization for Machine Learning (OPT 2019), co-located with NeurIPS 2019

In recent years, although the Alternating Direction Method of Multipliers (ADMM) has been empirically applied widely to many multi-convex applications, delivering an impressive performance in areas such as nonnegative matrix factorization and sparse dictionary learning, there remains a dearth of generic work on proposed ADMM with a convergence guarantee under mild conditions. In this paper, we propose a generic ADMM framework with multiple coupled variables in both objective and constraints. Convergence to a Nash point is proven with a sublinear convergence rate $o(1/k)$. Two important applications are discussed as special cases under our proposed ADMM framework. Extensive experiments on ten real-world datasets demonstrate the proposed framework's effectiveness, scalability, and convergence properties. We have released our code at \url{https://github.com/xianggebenben/miADMM}., Comment: Accepted by PAKDD 2022

Dry eye disease (DED) is a multifactorial disease of the tear film and ocular surface that causes symptoms of discomfort, visual disturbances and instability of the tear film. It is accompanied by an increase in tear film osmolarity and ocular surface inflammation. The global prevalence of Dry Eye Disease (DED) ranged from 5% to 50%, accompanied by the yearly increasing trend and younger onset. However, the condition remains incompletely understood and treated. DED symptom severity significantly correlates with deficits in physical, psychological, and social well-being, with the most severely affected patients reporting the worst scores for all aspects of health-related quality of life. DED symptoms and reduced visual function affect vision-dependent daily activities, with consequential negative impacts on social and physical functioning, workplace productivity, and quality of life. Furthermore, patients with DED report significantly higher rates of anxiety and depression compared with the general population, and significantly higher rates of sleep and mood disorders compared with other ocular conditions, which may exacerbate productivity, performance of personal tasks, and overall well-being. At present, there is no fully curable modality for DED. Artificial tear serves as a mainstay therapy for DED management. In fact, acupuncture has an outstanding effects in treating several diseases, including DED. Clinical study indicated that acupuncture could improve mid-term outcome compared to artificial tears. However, acupuncture is not recommended as a complementary therapy in the latest management guidelines of DED. We therefore think, it is necessary to conduct a comprehensive review of the latest research on the effects of acupuncture vs. artificial tears on DED treatment, with the aim to explore whether acupuncture deserves to be recommended as an adjunct therapy for DED.

Cloud storage security has drawn more and more concerns in Cloud Computing literatures. The Provable Data Possession (PDP) model has been defined for ensuring the integrity of files on untrusted storages. Since cloud storage is practically not only used for storing static data, the Dynamic Provable Data Possession (DPDP) model has been put forward, in which model, file content can be updated while the storage server is still able to prove the integrity and freshness of the file. In existing DPDP schemes, homomorphic MAC or signature is applied to achieve blockless verification in the course of integrity checking, whereas in the course of update verification, the verification protocol has to be performed for every single block update. In this paper, we develop a batch-update verifiable DPDP mode that verifies a batch of block updates at once to achieve efficiency. We construct our scheme by developing a variant authenticated 2–3 tree to enable data dynamics and batch-update verification. We also employ a BLS-like signature to enable public verification. The simulation shows that our batched scheme achieves great performance improvement on update verification.