Search

The advent of electric bikes, or ebikes, has significantly enhanced competitiveness of bike-sharing systems, providing benefits to both riders (comfort during uphill and long-distance rides), platforms (more profit), and the environment. Operating such a hybrid bike-sharing system, i.e., with both bikes and ebikes, in a competitive multi-platform market, can be challenging due to the complex and unpredictable interplay among heterogeneous market participants, which becomes more pronounced with the ebike varying battery, and dynamic demand. Most related research is predicated on the assumption of a monopoly market, which is not always the case: in worldwide capital-oriented markets, many firms will quickly imitate and join in rapidly developing fields for profits. Thus, this paper addresses platforms’ hybrid bike-sharing system operation problem with time-varying ebike pricing and rebalancing strategy in consideration of competition. We consider two docked hybrid bike-sharing platforms with charging stations at the site. Platforms utilize trucks for their own rebalancing operation including bike, ebike and mixed bike/ebike relocation tasks. We combine the Markov decision process (MDP) model with game theory, and establish the dual-platform MDP framework in which one mainstream platform and one competing platform optimize their profits by dynamic pricing and bike/ebike rebalancing based on highly dynamic and stochastic demand. Users’ choice is described by a modified nested logit model and the endogenous demand is generated. We develop the tailored double dueling deep Q-network for solving dynamic gaming. A series of experiments are conducted based on the real-world dataset in Shenzhen and several strategy combinations are compared. The results show the win–win situation where both platforms improve profits with a higher market ratio and demonstrate how to introduce and operate ebikes in the system by analyzing detailed strategies in different games. © 2023 Elsevier L

Autonomous vehicles (AVs), which can be fully controlled by remote/online operators, could be an extension of ride-sourcing services provided by transportation network companies (TNCs). Meanwhile, substitutive and complementary relationships between ride-sourcing and public transit could also help TNCs increase their profit under certain strategies. Unlike previous studies that generally ignored either AVs or interactions between public transit and ride-sourcing, we introduce AV-involved operational strategies into a multi-modal system. In this paper, we focus on transportation hubs in an urban area and let the TNC assign AVs to these hubs for serving only within that particular hub at a differentiated fare; meanwhile, passengers could choose among AVs, human-driven vehicles (HVs), combined modes with public transit and AVs or HVs, or other travel modes to fulfill their travel needs. We develop a mathematical model to investigate the impacts of such an operational strategy, and formulate an optimisation problem to maximise the TNC's profit and minimise total waiting time simultaneously by adjusting AV fare and fleet size. We further propose a comprehensive modeling framework with the analytical model, optimisation problem, calibration method, and heuristic algorithm, making it a general approach for different real-world scenarios. Following the framework, we conduct a case study based on real-world datasets of public transit and ride-sourcing services in Hangzhou, China. Different market schemes are analyzed and compared with the currently existing situation. The results demonstrate that by adopting this auxiliary-AV-oriented operational strategy, the TNC's profit and public transit ridership can both be increased, and passengers could enjoy a shorter waiting time for HV ride-sourcing trips. Moreover, the TNC is more inclined to allocate more AVs to hubs with large commute needs and uncongested traffic, leading to a high profit for the TNC and a short waiting time f

Modern large language models (LLMs) should generally benefit individuals from various cultural backgrounds around the world. However, most recent advanced generative evaluation benchmarks tailed for LLMs mainly focus on English. To this end, we introduce OMGEval, the first Open-source Multilingual Generative test set that can assess the capability of LLMs in different languages. For each language, OMGEval provides 804 open-ended questions, covering a wide range of important capabilities of LLMs, such as general knowledge, logical reasoning, and so on. Each question is rigorously verified by human annotators. Notably, to sufficiently reflect the compatibility of LLMs in different cultural backgrounds, we perform localization for each non-English language. Specifically, the current version of OMGEval includes 5 languages (i.e., Zh, Ru, Fr, Es, Ar). Following AlpacaEval, we employ GPT-4 as the adjudicator to automatically score different model outputs, which is shown closely related to human evaluation. We evaluate several representative multilingual LLMs on the proposed OMGEval, which we believe will provide a valuable reference for the community to further understand and improve the multilingual capability of LLMs. OMGEval is available at https://github.com/blcuicall/OMGEval.

Use-after-free (UAF) is a critical and prevalent problem in memory unsafe languages. While many solutions have been proposed, balancing security, run-time cost, and memory overhead (an impossible trinity) is hard. In this paper, we show one way to balance the trinity by passing more semantics about the heap object to the allocator for it to make informed allocation decisions. More specifically, we propose a new notion of thread-, context-, and flow-sensitive "type", SemaType, to capture the semantics and prototype a SemaType-based allocator that aims for the best trade-off amongst the impossible trinity. In SeMalloc, only heap objects allocated from the same call site and via the same function call stack can possibly share a virtual memory address, which effectively stops type-confusion attacks and makes UAF vulnerabilities harder to exploit. Through extensive empirical evaluation, we show that SeMalloc is realistic: (a) SeMalloc is effective in thwarting all real-world vulnerabilities we tested; (b) benchmark programs run even slightly faster with SeMalloc than the default heap allocator, at a memory overhead averaged from 41% to 84%; and (c) SeMalloc balances security and overhead strictly better than other closely related works., Comment: Accepted to ACM CCS 2024, camera-ready version under preparation

Recent advancements have introduced machine learning frameworks to enhance the Branch and Bound (B\&B) branching policies for solving Mixed Integer Linear Programming (MILP). These methods, primarily relying on imitation learning of Strong Branching, have shown superior performance. However, collecting expert samples for imitation learning, particularly for Strong Branching, is a time-consuming endeavor. To address this challenge, we propose \textbf{C}ontrastive Learning with \textbf{A}ugmented \textbf{M}ILPs for \textbf{Branch}ing (CAMBranch), a framework that generates Augmented MILPs (AMILPs) by applying variable shifting to limited expert data from their original MILPs. This approach enables the acquisition of a considerable number of labeled expert samples. CAMBranch leverages both MILPs and AMILPs for imitation learning and employs contrastive learning to enhance the model's ability to capture MILP features, thereby improving the quality of branching decisions. Experimental results demonstrate that CAMBranch, trained with only 10\% of the complete dataset, exhibits superior performance. Ablation studies further validate the effectiveness of our method.

Attacks on heap memory, encompassing memory overflow, double and invalid free, use-after-free (UAF), and various heap spraying techniques are ever-increasing. Existing entropy-based secure memory allocators provide statistical defenses against virtually all of these attack vectors. Although they claim protections against UAF attacks, their designs are not tailored to detect (failed) attempts. Consequently, to beat this entropy-based protection, an attacker can simply launch the same attack repeatedly with the potential use of heap spraying to further improve their chance of success. We introduce S2malloc, aiming to enhance UAF-attempt detection without compromising other security guarantees or introducing significant performance overhead. To achieve this, we use three innovative constructs in secure allocator design: free block canaries (FBC) to detect UAF attempts, random in-block offset (RIO) to stop the attacker from accurately overwriting the victim object, and random bag layout (RBL) to impede attackers from estimating the block size based on its address. We show that (a) by reserving 25% of the object size for the RIO offset, an 8-byte canary offers a 69% protection rate if the attacker reuses the same pointer and 96% protection rate if the attacker does not, against UAF exploitation attempts targeting a 64 bytes object, with equal or higher security guarantees against all other attacks; and (b) S2malloc is practical, with only a 2.8% run-time overhead on PARSEC and an 11.5% overhead on SPEC. Compared to state-of-the-art entropy-based allocators, S2malloc improves UAF-protection without incurring additional performance overhead. Compared to UAF-mitigating allocators, S2malloc trades off a minuscule probability of failed protection for significantly lower overhead., Comment: Accepted at DIMVA 2024, this is the extended version

Manual confirmation of static analysis reports is a daunting task. This is due to both the large number of warnings and the high density of false positives among them. Fuzzing techniques have been proposed to verify static analysis warnings. However, a major limitation is that fuzzing the whole project to reach all static analysis warnings is not feasible. This can take several days and exponential machine time to increase code coverage linearly. Therefore, we propose FuzzSlice, a novel framework that automatically prunes possible false positives among static analysis warnings. Unlike prior work that mostly focuses on confirming true positives among static analysis warnings, which requires end-to-end fuzzing, FuzzSlice focuses on ruling out potential false positives, which are the majority in static analysis reports. The key insight that we base our work on is that a warning that does not yield a crash when fuzzed at the function level in a given time budget is a possible false positive. To achieve this, FuzzSlice first aims to generate compilable code slices at the function level and then fuzzes these code slices instead of the entire binary. FuzzSlice is also unlikely to misclassify a true bug as a false positive because the crashing input can be reproduced by a fuzzer at the function level as well. We evaluate FuzzSlice on the Juliet synthetic dataset and real-world complex C projects. Our evaluation shows that the ground truth in the Juliet dataset had 864 false positives which were all detected by FuzzSlice. For the open-source repositories, we were able to get the developers from two of these open-source repositories to independently label these warnings. FuzzSlice automatically identifies 33 out of 53 false positives confirmed by developers in these two repositories. Thus FuzzSlice reduces false positives by 62.26% in the open-source repositories and by 100% in the Juliet dataset., Comment: The paper has been accepted for publication at ICSE 2024 (Research Track)

Human brains image complicated scenes when reading a novel. Replicating this imagination is one of the ultimate goals of AI-Generated Content (AIGC). However, current AIGC methods, such as score-based diffusion, are still deficient in terms of rapidity and efficiency. This deficiency is rooted in the difference between the brain and digital computers. Digital computers have physically separated storage and processing units, resulting in frequent data transfers during iterative calculations, incurring large time and energy overheads. This issue is further intensified by the conversion of inherently continuous and analog generation dynamics, which can be formulated by neural differential equations, into discrete and digital operations. Inspired by the brain, we propose a time-continuous and analog in-memory neural differential equation solver for score-based diffusion, employing emerging resistive memory. The integration of storage and computation within resistive memory synapses surmount the von Neumann bottleneck, benefiting the generative speed and energy efficiency. The closed-loop feedback integrator is time-continuous, analog, and compact, physically implementing an infinite-depth neural network. Moreover, the software-hardware co-design is intrinsically robust to analog noise. We experimentally validate our solution with 180 nm resistive memory in-memory computing macros. Demonstrating equivalent generative quality to the software baseline, our system achieved remarkable enhancements in generative speed for both unconditional and conditional generation tasks, by factors of 64.8 and 156.5, respectively. Moreover, it accomplished reductions in energy consumption by factors of 5.2 and 4.1. Our approach heralds a new horizon for hardware solutions in edge computing for generative AI applications.

Background: In the clinic, patients with endoprosthetic knee replacement for bone tumour generally experience loss of neuromuscular control and body balance, which consequently affects knee function and quality of life. Active physical exercise (APE) has been proven to be effective in rehabilitation. However, in the literature, no systematic study has been performed to make a general guideline of APE training on the patients. Methods: A total number of 528 patients who have had endoprosthetic knee replacement for tumour resection was collected from July 2009 to December 2016. After operation, the patients were randomly attributed to rehabilitation of either APE training or conventional continuous passive motion (CPM) for six months. The APE training was specific for patients with a bone tumour on the distal femur (APE-F) and the proximal tibia (APE-T), and the whole training was divided into three stages with different training items and training intensity which was progressively increased with limited intensity in terms of metabolic equivalent (MET) level for each stage. For each individual patient, the APE training intensity was personal based on personal basic MET. The patients were followed up for five years for regular assessments of functional knee recovery and quality of life. Findings: Both APE training and CPM training induced significant improvements in functional knee recovery and quality of life throughout the whole period of follow-up. Nevertheless, the improvements were more significant in patients following APE training than in that following CPM training (APE-F vs. CPM-F and APE-T vs. CPM-T). Additionally, the specific APE training induced significantly better functional knee recovery throughout the whole five-year of follow-up, and quality of life in the first six-month post-operation in patients of APE-F than in that of APE-T. In contrast, the CPM training induced significantly better improvements in both functional knee recovery and quality of lif

Topological insulators are emerging materials with insulating bulk and symmetry protected nontrivial surface states. One of the most fascinating transport behaviors in a topological insulator is the quantized anomalous Hall insulator, which has been observed in magnetic-topological-insulator-based devices. In this work, we report a successful doping of rare earth element Tb into Bi1.08Sb0.9Te2S topological insulator single crystals, in which the Tb moments are antiferromagnetically ordered below ∼10 K. Benefiting from the in-bulk-gap Fermi level, transport behavior dominant by the topological surface states is observed below ∼150 K. At low temperatures, strong Shubnikov-de Haas oscillations are observed, which exhibit 2D-like behavior. The topological insulator with long range magnetic ordering in rare earth doped Bi1.08Sb0.9Te2S single crystal provides an ideal platform for quantum transport studies and potential applications.

Objectives: To explore whether active physical exercise (APE) is more effective than conventional continuous passive motion (CPM) in improving functional knee recovery and quality of life without increasing the risk of complication in patients with endoprosthetic knee replacement for bone tumor. Methods: Six hundred and fourteen patients were enrolled and allocated to either APE or CPM for 6 months. APE was specific for patients with bone tumors on the distal femur (APE-F) and proximal tibia (APE-T), whereas CPM was similar to both types of patients (CPM-F and CPM-T). APE for both APE-F and APE-T patients was started on the second post-operation day, and CPM was initiated on the second post-procedure day in CPM-F patients, and in the seventh post-procedure week in CPM-T patients. The 6-month APE training consisted of three stages with training intensity progressively increasing though limited within a range of metabolic equivalent (MET) on each stage. Training intensity was individualized based on personal basic MET. The patients were followed up for 5 years for regular assessments of functional knee recovery, quality of life, and rate of complications. Results: APE-F and APE-T patients presented with significantly better results in functional knee recovery and quality of life in comparison with CPM-F and CPM-T patients, respectively. In addition, APE-T patients reached a similar level of quality of life as APE-F patients 6-month post-operation, whereas CPM-T patients reached a similar level of quality of life as CPM-F patients 3 years after the procedure. Nevertheless, APE training did not induce a higher incidence of complications than CPM training. Conclusions: APE training was more effective in improving functional knee recovery and quality of life without increasing the risk of complications than CPM training, thus, being strongly recommended to the patients for post-operation rehabilitation.

Angelica sinensis roots (Angelica roots) are rich in many bioactive compounds, including phthalides, coumarins, lignans, and terpenoids. However, the molecular bases for their biosynthesis are still poorly understood. Here, an improved chromosome-scale genome for A. sinensis var. Qinggui1 is reported, with a size of 2.16 Gb, contig N50 of 4.96 Mb and scaffold N50 of 198.27 Mb, covering 99.8% of the estimated genome. Additionally, by integrating genome sequencing, metabolomic profiling, and transcriptome analysis of normally growing and early-flowering Angelica roots that exhibit dramatically different metabolite profiles, the pathways and critical metabolic genes for the biosynthesis of these major bioactive components in Angelica roots have been deciphered. Multiomic analyses have also revealed the evolution and regulation of key metabolic genes for the biosynthesis of pharmaceutically bioactive components; in particular, TPSs for terpenoid volatiles, ACCs for malonyl CoA, PKSs for phthalide, and PTs for coumarin biosynthesis were expanded in the A. sinensis genome. These findings provide new insights into the biosynthesis of pharmaceutically important compounds in Angelica roots for exploration of synthetic biology and genetic improvement of herbal quality.

Meng Xu,1,* Ke Che,1,* Cong Wang,2,* Ya-Ru Chen,1 Meng-Yuan Chen,1 Guang-Lei Zhang,1 Hao Yu,3 Hao-Nan Xu,1 Ya-Bao Li,1 Ping Sheng,1 Hao Chen1 1College of Life and Health Sciences, Anhui Science and Technology University, Fengyang, 233100, People’s Republic of China; 2School of Environmental and Biological Engineering, Nanjing University of Science and Technology, Nanjing, 210094, People’s Republic of China; 3Bozhou University, Bozhou, 236800, People’s Republic of China*These authors contributed equally to this workCorrespondence: Hao Chen; Ping Sheng, Email chenhao@ahstu.edu.cn; astu100@126.comIntroduction: Cholestasis is a common liver disorder that currently has limited treatment options. Gardenia Iridoid Glucosides (GIG) have been found to possess various physiological activities, such as cholagogic, hypoglycemic, antibacterial, and anti-inflammatory effects. The objective of this study was to investigate the effects of GIG on bile acid enterohepatic circulation and explore the underlying mechanism in cholestatic rats.Methods: In order to identify key pathways associated with cholestasis, we conducted Gene Ontology (GO) Enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses. In vivo experiments were then performed on alpha-naphthylisothiocyanate (ANIT)-treated rats to assess the impact of GIG. We measured bile flow and various biomarkers including total bilirubin (TB), total bile acids (TBA), total cholesterol (TC), malondialdehyde (MDA), glutamic-pyruvic transaminase (GPT), glutamic oxaloacetic transaminase (GOT), and total superoxide dismutase (T-SOD) in the serum. We also examined the expression levels of bile salt export pump (BSEP), ATP-binding cassette subfamily B member 4 (ABCB4), far-nesoid X receptor (FXR), small heterodimer partner (SHP), cholesterol 7α-hydroxylase (CYP7A1), and sodium taurocholate cotransporting polypeptide (NTCP) in liver tissue. In vitro experiments were conducted on primary hepa

With the booming of transportation big data, plenty of spatial and temporal features become available for developing travel demand analysis models. However, existing machine learning based travel demand fitting models could face significant model complexity issues once there are multiple discrete/categorical independent variables. This paper embeds a non-local grouping (clustering) process into the tensor train decomposition (NLG-TTD) to characterize the higher-order and nonlinear relationship between categorical variables and travel demand. That is, the transportation dataset with multiple discrete/categorical independent variables and travel demand is treated as a multi-mode tensor, wherein levels of a discrete/categorical variable are regarded as indexes of a mode in the tensor, leading to a “regression on index”. The proposed NLG-TTD model divides the full tensor into non-local sub-tensors, clusters sub-tensors with similarity, conducts tensor train decomposition (TTD) within each cluster, and aggregates cluster-specific learning results. Compared with a conventional TTD that generally works on image completions, NLG-TTD can reveal non-Euclidean and non-linear patterns between categorical/discrete levels in independent variables. It is expected to discover similar travel demand patterns among categorical levels and provide high-level goodness-of-fit. For numerical studies, we use bike-sharing usage data and metro demand data in New York City (NYC), in which several widely-used machine learning models are trained for comparison. The results demonstrate the good performance of the proposed model. Moreover, we apply the NLG-TTD model to analyze and predict the impact of land use shifts on travel demand. © 2023 Elsevier Ltd

The data-driven Intelligent Transportation System (ITS) provides great support to travel decisions and system management but inevitably encounters the issue of data missing in monitoring systems. Hence, network-wide traffic state prediction and imputation is critical to recognizing the system level state of a transportation network. Abundant research works have adopted various approaches for traffic prediction and imputation. However, previous methods ignore the reliability analysis of the predicted/imputed traffic information. Thus, this study originally proposes an attentive graph neural process (AGNP) method for network-level short-term traffic speed prediction and imputation, simultaneously considering reliability. Firstly, the Gaussian process (GP) is used to model the observed traffic speed state. Such a stochastic process is further learned by the proposed AGNP method, which is utilized for inferring the congestion state on the remaining unobserved road segments. Data from a transportation network in Anhui Province, China, is used to conduct three experiments with increasing missing data ratio for model testing. Based on comparisons against other machine learning models, the results show that the proposed AGNP model can impute traffic networks and predict traffic speed with high-level performance. With the probabilistic confidence provided by the AGNP, reliability analysis is conducted both numerically and visually to show that the predicted distributions are beneficial to guide traffic control strategies and travel plans.

With the modern software and online platforms to collect massive amount of data, there is an increasing demand of applying causal inference methods at large scale when randomized experimentation is not viable. Weighting methods that directly incorporate covariate balancing have recently gained popularity for estimating causal effects in observational studies. These methods reduce the manual efforts required by researchers to iterate between propensity score modeling and balance checking until a satisfied covariate balance result. However, conventional solvers for determining weights lack the scalability to apply such methods on large scale datasets in companies like Snap Inc. To address the limitations and improve computational efficiency, in this paper we present scalable algorithms, DistEB and DistMS, for two balancing approaches: entropy balancing and MicroSynth. The solvers have linear time complexity and can be conveniently implemented in distributed computing frameworks such as Spark, Hive, etc. We study the properties of balancing approaches at different scales up to 1 million treated units and 487 covariates. We find that with larger sample size, both bias and variance in the causal effect estimation are significantly reduced. The results emphasize the importance of applying balancing approaches on large scale datasets. We combine the balancing approach with a synthetic control framework and deploy an end-to-end system for causal impact estimation at Snap Inc., Comment: KDD '23: Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining

Traffic flow prediction is an essential component in intelligent transportation systems. Recently, there has been a notable trend in applying machine learning models, especially deep learning, for network-wide traffic prediction. However, existing studies have limitations on model interpretability, model generalization, and over-reliance on image data processing or fine-designed deep learning structures for extracting traffic attributes. This paper attempts to tackle these limitations by proposing a Bayesian clustering ensemble Gaussian process (BCEGP) model for network-wide traffic flow clustering and prediction. The model utilizes a subset-based Dirichlet process mixture (SDPM) model to conduct a hard clustering among input data; then, within each cluster, it adopts the Gaussian Process (GP) to learn the probability relationship between inputs and outputs. During the prediction phase, the model conducts a soft clustering of the input as weights, and makes predictions via a weighted average of GPs’ outputs. The merits of the BCEGP model include: (a) data with similar spatial–temporal patterns are clustered, which helps understand traffic dynamics in a non-Euclidean and non-graphical manner that enhances information extracting for model development; (b) GPs provide analytically trackable functions/gradients of predicted traffic flows with features and reveal variances of predicted traffic flow, enhancing model applicability and interpretability to some extent; (c) the model incorporates an ensemble learning framework that achieves great generalization performance as good as deep learning models; (d) the subset-based clustering and cluster-based GP learning are conducted parallelly, and thus vastly accelerate the training efficiency compared with conventional GPs (but slower than deep learning models). We test the performance of the proposed model based on both synthesized and real-world datasets. For comparison, several widely used machine learning and deep learning

With the rise of sharing economy, bike-sharing systems (BSSs) have gained heated attention, and their operations require accurate prediction of bike usage. Although many deep learning methods have been exploited to predict bike usage, they generally provide point predictions of average bike usage, neglecting the stochasticity in BSSs. Due to the analytically explainable properties and linear computational costs with respect to data size, neural processes (NPs) have recently attracted increasing interest. An NP model learns a Gaussian process (GP) by mapping the input–output observations to a probabilistic distribution over functions. Each function is a distribution of the outputs given an input, conditioned on the arbitrary size of observed data. NPs provide probabilistic confidence in predicted results, which overcomes the point prediction issue faced by other models and provides insights for operational strategies in stochastic scenarios. This paper originally proposes a transformer-encoder-based NP (TENP) model to fit the distribution of bike usage in BSSs. To the best of our knowledge, this work is among the first to incorporate transformer encoders into NPs, enhancing the capability of extracting relevant information in a targeted manner. Based on the Citi Bike datasets in New York City, the TENP method is adopted in a multi-task learning task that simultaneously fits the number of pickups and returns. The proposed TENP model outperforms the conventional NP method and its extensions and prevalent machine learning models in terms of prediction accuracy. Armed with the probabilistic confidence provided by the TENP, reliability analysis is conducted, and thoughtful guidance is provided for bike-sharing operations, such as dynamic bike rebalancing. © 2023 Elsevier Ltd

Matching and pickup processes are core features of ride-sourcing services. Previous studies have adopted abundant analytical models to depict the two processes and obtain operational insights; while the goodness of fit between models and data was dismissed. To simultaneously consider the fitness between models and data and analytically tractable formations, we propose a data-driven approach based on the additive Gaussian Process Model (AGPM) for ride-sourcing market modelling. The framework is tested based on real-world data collected in Hangzhou, China. We fit analytical models, machine learning models, and AGPMs, in which the number of matches or pickups are used as outputs and spatial, temporal, demand, and supply covariates are utilized as inputs. The results demonstrate the advantages of AGPMs in recovering the two processes in terms of estimation accuracy. Furthermore, we illustrate the modelling power of AGPM by utilizing the trained model to design and estimate idle vehicle relocation strategies. © 2022 Hong Kong Society for Transportation Studies Limited.

Construction waste hauling trucks (or `slag trucks') are among the most commonly seen heavy-duty diesel vehicles in urban streets, which not only produce significant carbon, NO$_{\textbf{x}}$ and PM$_{\textbf{2.5}}$ emissions but are also a major source of on-road and on-site fugitive dust. Slag trucks are subject to a series of spatial and temporal access restrictions by local traffic and environmental policies. This paper addresses the practical problem of predicting levels of slag truck activity at a city scale during heavy pollution episodes, such that environmental law enforcement units can take timely and proactive measures against localized truck aggregation. A deep ensemble learning framework (coined AI-Truck) is designed, which employs a soft vote integrator that utilizes Bi-LSTM, TCN, STGCN, and PDFormer as base classifiers. AI-Truck employs a combination of downsampling and weighted loss is employed to address sample imbalance, and utilizes truck trajectories to extract more accurate and effective geographic features. The framework was deployed for truck activity prediction at a resolution of 1km$\times$1km$\times$0.5h, in a 255 km$^{\textbf{2}}$ area in Chengdu, China. As a classifier, AI-Truck achieves a macro F1 of 0.747 in predicting levels of slag truck activity for 0.5-h prediction time length, and enables personnel to spot high-activity locations 1.5 hrs ahead with over 80\% accuracy., Comment: 11 pages, 15 figures

Ebike-sharing (electric bicycle-sharing) systems are gaining popularity as ebikes provide riders with transportation convenience when people have limited accessibility to other travel modes. Compared to traditional bike-sharing systems, ebike-sharing systems are more complicated as the platform needs to handle battery recharging issues as well as the imbalance between supply and demand. However, previous studies have not discussed how to address the two issues effectively. In this paper, we consider a dockless ebike-sharing system with removable ebike batteries and introduce vans to such a system to solve recharging and rebalancing problems simultaneously. In other words, during the operational horizon, van drivers can choose among rebalancing tasks, battery swapping tasks, and half-rebalancing-half-swapping tasks. This paper utilizes the Markov decision process to depict the highly dynamic ebike-sharing system with a platform player (agent) and a representative van driver player (agent). In such a system, van drivers choose their tasks to maximize their income, and the platform allocates spatiotemporal subsidies with predefined subsidy amounts to incentivize van drivers and optimize the profit. To efficiently solve the dynamic optimization problem with mixed agents, we apply the dueling double deep Q-network method which is an advanced reinforcement learning approach. We conduct numerical studies based on a real-world dataset in New York City, and evaluate the performance of the proposed operational services under different schemes. Our results show that the proposed van-based services for battery swapping and rebalancing could help increase the platform's profit and van drivers' earnings, and improve system performance. Additionally, it is also proved that the platform, van drivers, and the overall ebike-sharing system all benefit from spatiotemporal subsidies. © 2022 Elsevier Ltd

Patients with postoperative tumor-associated bone defects need both chemotherapy and bone repair in the shifted disease states, but chemotherapy resistance, severe side-effects, and poor bone regeneration result in tumor recurrence and therapy failure. Herein, we develop an injectable and tumor microenvironment (TME)-modulated hydrogel composed of bio-responsive drug-loaded mesoporous bioactive glass nanoparticles (MBGN), gelatin (Gel), and oxidized chondroitin sulfate (OCS) for tumor-associated bone defects closed-loop management. The TME-modulated hydrogels can deplete protons and glutathione (GSH) within residual tumor tissues with sustained drug responsive-release capability, therefore interfering TME and overcoming cancer resistance, and also reducing off-target effects. Mechanistically, in the early state of treatment, acidic environment of TME will break the Schiff's base bond network of hydrogels, thus deliver drug-loaded nanoparticles into tumor cells. Subsequently, the pH-responsive amide bonds, GSH-responsive disulfide bonds and acidity-degradable nanocarriers of drug-loaded nanoparticles can induce ROS stress and mitochondrial-related damage, activating the p16-CYLD tumor suppressor pathway to induce apoptosis and inhibit metastasis. In the late state of treatment, the hydrogels shifted into regenerative scaffolds to exert long-lasting osteogenesis effects. Overall, the TME-modulated hydrogels can orchestrate steady bone repair and avoid tumor recurrence, achieving closed-loop management of postoperative tumor-associated bone defects.

Jing Mao,1 Huiping Guo,2 Juan Wang,1 Ying Li,1 Meng Xu1 1Department of Reproductive Medicine, the First Affiliated Hospital of Henan University of Chinese Medicine, Zhengzhou, People’s Republic of China; 2Nursing College, Henan University of Chinese Medicine, Zhengzhou, People’s Republic of ChinaCorrespondence: Jing Mao, Department of Reproductive Medicine, the First Affiliated Hospital of Henan University of Chinese Medicine, No. 19 Renmin Road, Jinshui District, Zhengzhou, 450000, People’s Republic of China, Tel +86-18703657086, Email jingmao_99@yeah.netObjective: To investigate the psychosocial factors, including personality characteristics, coping style, social support and alexithymia of infertile patients, and explore the relevant factors regarding the medical coping style choice of infertile patients in central China.Methods: Sixty infertile patients in the reproductive medicine centre in our hospital were investigated using a self-made general situation questionnaire; internal control, powerful others and opportunities scale; social support rating scale; and medical coping questionnaire, and the data was analysed.Results: Of the 60 infertile patients, 57 (95%) were female, with an average age of 27.85 ± 14.52 years old, and 42 (70%) had secondary infertility. The score of the face dimension was lower than the norm (19.17 ± 2.18 vs 19.43 ± 3.82, P = 0.012), and those of the avoidance (15.98 ± 2.43 vs 14.41 ± 2. 94, P = 0.030) and yield (9.48 ± 3.19 vs 8.83 ± 3.19, P = 0.012) dimensions were higher than the norm. The multiple linear regression analysis showed that internal control and the utilisation of support were influencing factors for the coping style (R2 = 0.271, adjusted R2 = 0.144, F = 9.827, P < 0.001), and internal control was the influencing factor for the avoidance coping style (R2 = 0.024, adjusted R2 = 0.021, F = 3.363, P = 0.014). The education level, opportunity and utilisation of support factors influenced the yield dimension (R

Asphalt aging often leads to rapid degradation of road performance, which seriously affects the service life of asphalt pavement. Exploring the influence of asphalt oil sources, asphalt grades, and filler types on asphalt microcharacteristics in the asphalt aging process can provide an essential reference to guide asphalt pavement maintenance. In this study, we selected seven kinds of asphalt and three fillers commonly used in China for research. The pressurized aging vessel (PAV) and homemade ultraviolet (UV) aging equipment were used to perform thermo-oxidative aging and UV aging tests, respectively, of asphalt. The microcharacteristics of asphalt before and after aging were analyzed via attenuated total reflectance fourier transformation infrared spectroscopy and nuclear magnetic resonance 1H spectroscopy. The results show that the oil source of asphalt exerted the most significant influence on the microcharacteristics of the aged asphalt, while the effect of the asphalt grade was relatively limited. The addition of fillers did not affect the aging mechanism of asphalt. UV and PAV aging generated apparent differences in the changes in the aged asphalt microstructure., Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public., Pavement Engineering

Accurate localization ability is fundamental in autonomous driving. Traditional visual localization frameworks approach the semantic map-matching problem with geometric models, which rely on complex parameter tuning and thus hinder large-scale deployment. In this paper, we propose BEV-Locator: an end-to-end visual semantic localization neural network using multi-view camera images. Specifically, a visual BEV (Birds-Eye-View) encoder extracts and flattens the multi-view images into BEV space. While the semantic map features are structurally embedded as map queries sequence. Then a cross-model transformer associates the BEV features and semantic map queries. The localization information of ego-car is recursively queried out by cross-attention modules. Finally, the ego pose can be inferred by decoding the transformer outputs. We evaluate the proposed method in large-scale nuScenes and Qcraft datasets. The experimental results show that the BEV-locator is capable to estimate the vehicle poses under versatile scenarios, which effectively associates the cross-model information from multi-view images and global semantic maps. The experiments report satisfactory accuracy with mean absolute errors of 0.052m, 0.135m and 0.251$^\circ$ in lateral, longitudinal translation and heading angle degree.

Camera, and associated with its objects within the field of view, localization could benefit many computer vision fields, such as autonomous driving, robot navigation, and augmented reality (AR). In this survey, we first introduce specific application areas and the evaluation metrics for camera localization pose according to different sub-tasks (learning-based 2D-2D task, feature-based 2D-3D task, and 3D-3D task). Then, we review common methods for structure-based camera pose estimation approaches, absolute pose regression and relative pose regression approaches by critically modelling the methods to inspire further improvements in their algorithms such as loss functions, neural network structures. Furthermore, we summarise what are the popular datasets used for camera localization and compare the quantitative and qualitative results of these methods with detailed performance metrics. Finally, we discuss future research possibilities and applications.

Ride-sourcing services have become an integral part of urban transportation systems. In a cooperative relationship with the platform, ride-sourcing drivers have the flexibility to work according to their habits or preferences, which could constitute heterogeneous working pattern(s) among drivers. Based on real-world ride-sourcing data, we analyze drivers' working pattern(s) via their work slices (WSs). A WS is defined as a time window during which a driver is continuously online to take ride-sourcing orders. We use the Gaussian mixture model to classify the WSs according to the start time/start place attributes and obtain 9 clusters with significant spatiotemporal differences. It is found that WSs starting from rural areas that are closed to urban areas show no significant spatiotemporal grouping and mediocre performance in terms of fare income and fare-cost ratio, while WSs starting from distant rural areas show the worst performance; WSs starting from urban/suburban areas reflect more diverse spatiotemporal groupings and significant differences in attributes of delivery time and fares earned. We also analyze drivers' working pattern(s) by summarizing their habitual WS clusters. Findings include that drivers with suburban-dominant/rural-dominant patterns have a disadvantage in working efficiency in terms of the shorter order distance and longer cruising time. If geometrical distance allows, they can improve their income by switching to urban WS clusters occasionally. However, frequently switching between WS clusters can lead to a decrease in fares income. According to our results, the platform can introduce differentiated policies, including subsidies and order-matching algorithms, to better handle the spatiotemporal inequality, ultimately improving the service performance and overall benefits. © 2022 Elsevier Ltd

Fitting and modeling spatial-temporal processes are essential research topics in transportation studies. Recently, due to the analytically tractable formulation and good fitting accuracy, the Gaussian Process (GP) is becoming increasingly preferable in fitting transportation processes. However, conventional GPs are inapplicable for large-scale problems due to computational issues. Moreover, they dismiss physics laws when conducting multi-output fitting tasks. This paper proposes a physics regularized multi-output grid Gaussian Process Model (PRMGGP) model for fast and multi-output fitting of large-scale spatial-temporal processes in transportation systems. The PRMGGP model adopts a grid input structure to capture inherent spatial-temporal correlations in the fitting process, takes advantage of the Kronecker algebra to notably accelerate the computation speed, and utilizes a shadow GP to incorporate physics laws of the process. Model training and predictive algorithms are developed coordinately and are tested via synthetic datasets. Furthermore, we apply the proposed model and other widely used machine learning models to fit the numbers of pickups, returns, and idle bikes of a large-scale bike-sharing system based on Citi Bike data from New York City. The results demonstrate the computational efficiency, interpretable results, and the prediction accuracy of the PRMGGP model, which can be a promising methodology for modeling multi-output processes in transportation systems.

Meng-Nan Xu,1,&ast; Qian Wang,1,&ast; Min Wang,1 Yuan Xu,1 Si-Ming Yuan1,2 1Department of Plastic Surgery, Jinling Hospital, Medical School of Nanjing University, Nanjing, Jiangsu, 210002, People’s Republic of China; 2Department of Plastic Surgery, Jinling Hospital, Nanjing School of Clinical Medicine, Southern Medical University, Nanjing, Jiangsu, 210002, People’s Republic of China&ast;These authors contributed equally to this workCorrespondence: Si-Ming Yuan, Department of Plastic Surgery, Jinling Hospital, Medical School of Nanjing University, Nanjing, Jiangsu, 210002, People’s Republic of China, Email yuansm@163.comBackground: Port wine stain (PWS) is a congenital skin lesion involving capillary malformations. Most PWS lesions will gradually become hypertrophic and appear nodular in contour. Current research shows that rapamycin, an mTOR inhibitor, is probably a promising adjunctive therapy for PWS, which suggests that the mTOR signaling pathway may play an important role in its pathological process.Methods: From January 2013 to January 2019, 13 samples were obtained during the surgical excision. Each sample was divided into 3 parts according to the type of lesion, namely, the flat, hypertrophic and nodular lesions. Pathologic structures of each type were observed under the microscope after HE staining. The expression of mTORC1, p70S6, p-p70S6, eIF4EBP1 and p-eIF4EBP1 was examined by immunohistochemical staining and western blotting. The location of the expression of mTORC1, p-p70S6 and p-elF4EBP1 was further detected by immunofluorescence staining.Results: Large amounts of dilated and malformed vessels were observed in all types of PWS lesions. Abundant hyperplastic hair follicles/glands were shown in the hypertrophic or nodular lesions. Phosphorylation level of p70S6 and elF4EBP1 in PWS was significantly higher than those in normal skin and increased accordingly in the progression of PWS. Activated molecules in mTOR signaling pathway were mos

Hepatocellular carcinoma (HCC) is the most common type of liver cancer and has limited treatment options. Snail family transcriptional repressor 1 (SNAI1) is a master regulator of epithelial-mesenchymal transition (EMT) and has been implicated in HCC initiation and progression. However, the precise role of SNAI1 and the way it contributes to hepatocarcinogenesis have not been investigated in depth, especially in vivo. Here, we analyzed the functional relevance of SNAI1 in promoting hepatocarcinogenesis in the context of the AKT/c-Met-driven mouse liver tumor model (AKT/c-Met/SNAI1). Overexpression of SNAI1 did not accelerate AKT/c-Met-induced HCC development or induce metastasis in mice. Elevated SNAI1 expression rather led to the formation of cholangiocellular (CCA) lesions in the mouse liver, a phenotype that was paralleled by increased activation of Yap and Notch. Ablation of Yap strongly inhibited AKT/c-Met/SNAI-induced HCC and CCA development, whereas inhibition of the Notch pathway specifically blocked the CCA-like phenotype in mice. Intriguingly, overexpression of SNAI1 failed to induce EMT, indicated by strong E-cadherin expression and lack of vimentin expression by AKT/c-Met/SNAI tumor cells. SNAI1 mRNA levels strongly correlated with the expression of CCA markers, including SOX9, CK19, and EPCAM, but not with EMT markers such as E-CADHERIN and ZO-1, in human HCC samples. Overall, our findings suggest SNAI1 regulates the CCA-like phenotype in hepatocarcinogenesis via regulation of Yap and Notch. SIGNIFICANCE: These findings report a new function of SNAI1 to promote cholangiocellular transdifferentiation instead of epithelial-mesenchymal transition in hepatocellular carcinoma.

ObjectiveHepatocellular carcinoma (HCC) is the most common type of primary liver cancer with limited treatment options. Cabozantinib, an orally bioavailable multikinase inhibitor is now approved by Food and Drug Administration (FDA) for HCC patients. We evaluated the therapeutic efficacy of cabozantinib, either alone or in combination, in vitro and in vivo.DesignHuman HCC cell lines and HCC mouse models were used to assess the therapeutic efficacy and targeted molecular pathways of cabozantinib, either alone or in combination with the pan-mTOR inhibitor MLN0128 or the checkpoint inhibitor anti-PD-L1 antibody.ResultsCabozantinib treatment led to stable disease in c-Met/β-catenin and Akt/c-Met mouse HCC while possessing limited efficacy on Akt/Ras and c-Myc liver tumours. Importantly, cabozantinib effectively inhibited c-MET and ERK activity, leading to decreased PKM2 and increased p21 expression in HCC cells and in c-Met/β-catenin and Akt/c-Met HCC. However, cabozantinib was ineffective in inhibiting the Akt/mTOR cascade. Intriguingly, a strong inhibition of angiogenesis by cabozantinib occurred regardless of the oncogenic drivers. However, cabozantinib had limited impact on other tumour microenvironment parameters, including tumour infiltrating T cells, and did not induce programmed death-ligand 1 (PD-L1) expression. Combining cabozantinib with MLN0128 led to tumour regression in c-Met/β-catenin mice. In contrast, combined treatment with cabozantinib and the checkpoint inhibitor anti-PD-L1 antibody did not provide any additional therapeutic benefit in the four mouse HCC models tested.Conclusionc-MET/ERK/p21/PKM2 cascade and VEGFR2-induced angiogenesis are the primary targets of cabozantinib in HCC treatment. Combination therapies with cabozantinib and mTOR inhibitors may be effective against human HCC.

BACKGROUND: Adsorption has been widely applied in wastewater treatment for heavy metal removal. Due to the high cost of adsorbent materials, various types of materials including agricultural, industrial wastes and naturally abundant materials have emerged as low-cost adsorbents in recent years. However, enhancement processes need to be developed to improve the adsorption capability of the raw materials. It is still a challenge to find an economic material with high adsorption capacity. This paper demonstrates a strategy to activate a recycled waste material using a simple activating process, while significantly improving the adsorption capacity of cadmium from wastewater. RESULTS: The raw material feedstock is the non-metallic aluminosilicate material obtained from the recycling of waste printed circuit boards (PCBs). By a simple thermal alkali activation process using potassium hydroxide as activation agent, the product material shows a strong capability of cadmium removal from effluent containing a high adsorption capacity of 2.12 mmol g−1. Initial metal concentration, dosage and pH effect on the adsorption isotherm and kinetics have been investigated. The experimental results include full characterization tests on this waste PCB-derived adsorbent, and the in situ measurement of cadmium, calcium and potassium ions, since all are involved in the exchange sorption mechanism. CONCLUSION: By combining kinetic modelling analysis and X-ray photoelectron spectroscopy, the novel adsorption mechanism has been explained and quantified as a dual ion-exchange mechanism with the millimoles of cadmium exchanged mass balancing with the millimoles of calcium and 0.5 mmol potassium going into solution. © 2021 Society of Chemical Industry (SCI). © 2021 Society of Chemical Industry (SCI).

Background & aimsFocal adhesion kinase (FAK) is a non-receptor tyrosine kinase that is upregulated in many tumor types and is a promising target for cancer therapy. Herein, we elucidated the functional role of FAK in intrahepatic cholangiocarcinoma (iCCA) development and progression.MethodsExpression levels and activation status of FAK were determined in human iCCA samples. The functional contribution of FAK to Akt/YAP murine iCCA initiation and progression was investigated using conditional Fak knockout mice and constitutive Cre or inducible Cre mice, respectively. The oncogenic potential of FAK was further examined via overexpression of FAK in mice. In vitro cell line studies and in vivo drug treatment were applied to address the therapeutic potential of targeting FAK for iCCA treatment.ResultsFAK was ubiquitously upregulated and activated in iCCA lesions. Ablation of FAK strongly delayed Akt/YAP-driven mouse iCCA initiation. FAK overexpression synergized with activated AKT to promote iCCA development and accelerated Akt/Jag1-driven cholangiocarcinogenesis. Mechanistically, FAK was required for YAP(Y357) phosphorylation, supporting the role of FAK as a central YAP regulator in iCCA. Significantly, ablation of FAK after Akt/YAP-dependent iCCA formation strongly suppressed tumor progression in mice. Furthermore, a remarkable iCCA growth reduction was achieved when a FAK inhibitor and palbociclib, a CDK4/6 inhibitor, were administered simultaneously in human iCCA cell lines and Akt/YAP mice.ConclusionsFAK activation contributes to the initiation and progression of iCCA by inducing the YAP proto-oncogene. Targeting FAK, either alone or in combination with anti-CDK4/6 inhibitors, may be an effective strategy for iCCA treatment.Lay summaryWe found that the protein FAK (focal adhesion kinase) is upregulated and activated in human and mouse intrahepatic cholangiocarcinoma samples. FAK promotes intrahepatic cholangiocarcinoma development, whereas deletion of

With the rapid development of deep learning technology and improvement in computing capability, deep learning has been widely used in the field of hyperspectral image (HSI) classification. In general, deep learning models often contain many trainable parameters and require a massive number of labeled samples to achieve optimal performance. However, in regard to HSI classification, a large number of labeled samples is generally difficult to acquire due to the difficulty and time-consuming nature of manual labeling. Therefore, many research works focus on building a deep learning model for HSI classification with few labeled samples. In this article, we concentrate on this topic and provide a systematic review of the relevant literature. Specifically, the contributions of this paper are twofold. First, the research progress of related methods is categorized according to the learning paradigm, including transfer learning, active learning and few-shot learning. Second, a number of experiments with various state-of-the-art approaches has been carried out, and the results are summarized to reveal the potential research directions. More importantly, it is notable that although there is a vast gap between deep learning models (that usually need sufficient labeled samples) and the HSI scenario with few labeled samples, the issues of small-sample sets can be well characterized by fusion of deep learning methods and related techniques, such as transfer learning and a lightweight model. For reproducibility, the source codes of the methods assessed in the paper can be found at https://github.com/ShuGuoJ/HSI-Classification.git.

The Move Prover (MVP) is a formal verifier for smart contracts written in the Move programming language. MVP has an expressive specification language, and is fast and reliable enough that it can be run routinely by developers and in integration testing in a few minutes. Besides the simplicity of smart contracts and the Move language, three transformations are responsible for the practicality of MVP: (1) an alias-free memory model, (2) fine-grained invariant checking, and (3) monomorphization. The entirety of the Move code for the Diem blockchain has been extensively specified and can be completely verified by MVP in a few minutes. Changes in the Diem framework must be successfully verified before being integrated into the open source repository on GitHub.

Ternary I-III-VI2 nanocrystals (NCs), such as AgInS2 and CuInS2, are garnering interest as heavy-metal-free materials for photovoltaics, luminescent solar concentrators, LEDs, and bioimaging. The origin of the emission and absorption properties in this class of NCs is still a subject of debate. Recent theoretical and experimental studies revealed that the characteristic Stokes-shifted and long-lived luminescence of stoichiometric CuInS2 NCs arises from the detailed structure of the valence band featuring two sublevels with different parity. The same valence band substructure is predicted to occur in AgInS2 NCs, yet no experimental confirmation is available to date. Here, we use complementary spectroscopic, spectro-electrochemical, and magneto-optical investigations as a function of temperature to investigate the band structure and the excitonic recombination mechanisms in stoichiometric AgInS2 NCs. Transient transmission measurements reveal the signatures of two subbands with opposite parity, and photoluminescence studies at cryogenic temperatures evidence a dark state emission due to enhanced exchange interaction, consistent with the behavior of stoichiometric CuInS2 NCs. Lowering the temperature as well as applying reducing electrochemical potentials further suppress electron trapping, which represents the main nonradiative channel for exciton decay, leading to nearly 100% emission efficiency.

Image super-resolution (SR) methods can generate remote sensing images with high spatial resolution without increasing the cost, thereby providing a feasible way to acquire high-resolution remote sensing images, which are difficult to obtain due to the high cost of acquisition equipment and complex weather. Clearly, image super-resolution is a severe ill-posed problem. Fortunately, with the development of deep learning, the powerful fitting ability of deep neural networks has solved this problem to some extent. In this paper, we propose a network based on the generative adversarial network (GAN) to generate high resolution remote sensing images, named the multi-attention generative adversarial network (MA-GAN). We first designed a GAN-based framework for the image SR task. The core to accomplishing the SR task is the image generator with post-upsampling that we designed. The main body of the generator contains two blocks; one is the pyramidal convolution in the residual-dense block (PCRDB), and the other is the attention-based upsample (AUP) block. The attentioned pyramidal convolution (AttPConv) in the PCRDB block is a module that combines multi-scale convolution and channel attention to automatically learn and adjust the scaling of the residuals for better results. The AUP block is a module that combines pixel attention (PA) to perform arbitrary multiples of upsampling. These two blocks work together to help generate better quality images. For the loss function, we design a loss function based on pixel loss and introduce both adversarial loss and feature loss to guide the generator learning. We have compared our method with several state-of-the-art methods on a remote sensing scene image dataset, and the experimental results consistently demonstrate the effectiveness of the proposed MA-GAN.

The first chapter of my dissertation uses a competition framework to reexamine the U.S. railway expansion from 1870 to 1890. Do reductions in trade costs increase the income levels of trading regions? The answer of most theoretical trade models is yes, and these models are used by policy-makers to justify investments in new transportation infrastructure to reduce trade costs. I reexamine this question by investigating the impact of the U.S. railway expansion from 1870 to 1890 on county economic growth. My main insight is that while the construction of a new transportation network can benefit a location by increasing producers' access to markets, it also can hurt a location by giving other locations a competitive edge. I derive and test empirically testable implications of the joint effects of market access and competition on economic outcomes from a multisector trade model. I find that on average more than 50% of the benefit of market access is offset by the negative effects of competition. Overall, I find that the expansion of railways increased agricultural land values by 20% to 40%, thus implying that railways explain 1% to 2% of 1890 GNP. My estimates of the impact of railways on American economic growth are lower than those of Donaldson and Hornbeck (2016) and closer to those of Fogel (1964).In the second chapter I study the effects of sector-specific productivity shocks on the distribution of production. The general-equilibrium trade model implies that the production in a certain location depends not only on its own productivity shocks but also on productivity shocks in competing locations. I derive several testable implications of the joint effects of one location's own productivity shock and productivity shocks in competing locations on the location's production. I test these implications by investigating the impact of the boll weevil in the early 20th century United States on the cotton belt's cotton acreage. I find (1) when few lands in the cotton belt wer

Focusing on the price issue in shared parking, two auction-based pricing strategies are designed in a double environment covering parking suppliers and demanders. A novel uniform price strategy (UPS) is proposed to set unique transaction prices for winning participants, it is in contrast with the differential price strategy (DPS), which sets differential transaction prices. The DPS and UPS are compared from the perspective of parking platform's payoff and participants' utilities theoretically and numerically. It shows that the DPS is superior to the UPS in terms of parking platform's payoff whilst the UPS outperforms the DPS in terms of participants' utilities.

BackgroundThere is significant interindividual variability in the rate of aortic stenosis (AS) progression that is not accounted for in the current surveillance algorithms. We sought to examine the association between changes in peak aortic jet velocity (Vmax) and mean gradient (MG) among patients with mild or moderate AS and risk of progression to severe disease.MethodsAdult subjects referred for echocardiography at a single academic referral center with a diagnosis of mild or moderate AS and ≥2 additional surveillance echocardiograms were included in the study. Changes in Vmax and MG between the first two echocardiograms were indexed to time and tested for association with future progression to severe AS.ResultsAmong three hundred and sixty-four subjects, the median time between first and second echocardiograms was 1.3 years and initial changes in Vmax and MG indexed to time were +0.16 m/s per year and +1.44 mmHg per year, respectively. Fifty-three (15%) and fifty-six (15%) subjects progressed to severe AS defined by Vmax and MG, respectively. In multivariable logistic regression, initial increase in Vmax (OR = 4.19, 95% CI 1.93-9.10, p < 0.001) and initial increase in MG (OR = 1.12, 95% CI 1.06-1.18, p < 0.001) were associated with progression to severe AS.ConclusionsInitial changes in Vmax and MG among patients with mild or moderate AS are strongly associated with risk of progression to severe AS and may help guide individualized surveillance strategies.

Everyday life in urban public space means living amongst people unknown to one another. As part of the broader convivial turn within the study of everyday urban life (Wise & Noble, 2016), this article examines outdoor public ice rinks as spaces for encounter between strangers. With data drawn from 100 hours of naturalistic and participant observation at free and accessible outdoor public non-hockey ice rinks in two Canadian cities, we show how ‘rink life’ is animated by a shared everyday ethic of public sociability, with strangers regularly engaging in fleeting moments of sociable interaction. At first glance, researching the outdoor public ice rink may seem frivolous, but in treating it seriously as a public space we find it to be threaded through with an ethos of interactional equality, reciprocal respect, and mutual support. We argue that the shared everyday ethic of public sociability that characterizes the rinks that we observed is a function of the (1) public and (2) personal materiality required for skating; (3) the emergence of on ice norms; (4) generalized trust amongst users; (5) ambiguities of socio-spatial differentiation by skill; and (6) flattened social hierarchies, or what we call the quotidian carnivalesque. Our data and analysis suggest that by drawing together different generations and levels of ability, this distinct public space facilitates social interactions between strangers, and so provides insights relevant to planners, policy makers and practitioners.

Cholestatic liver injury may lead to a series of hepatobiliary syndromes, which can progress to cirrhosis and impaired liver regeneration, eventually resulting in liver-related death. Mammalian target of rapamycin complex 2 (mTORC2) is a major regulator of liver metabolism and tumor development. However, the role of mTORC2 signaling in cholestatic liver injury has not been characterized to date. In this study, we generated liver-specific Rictor knockout mice to block the mTORC2 signaling pathway. Mice were treated with 3,5-diethoxycarbonyl-1,4-dihydrocollidine (DDC) to induce cholestatic liver injury. DDC feeding induced cholestatic liver injury and ductular reaction as well as activation of the mTORC2/Akt signaling pathway in wild-type mice. Loss of mTORC2 led to significantly decreased oval cell expansion after DDC feeding. Mechanistically, this phenotype was independent of mTORC1/fatty acid synthase cascade (Fasn) or yes-associated protein (Yap) signaling. Notch pathway was instead strongly inhibited during DDC-induced cholestatic liver injury in liver-specific Rictor knockout mice. Furthermore, mTORC2 deficiency in adult hepatocytes did not inhibit ductular reaction in this cholestatic live injury mouse model. Our results indicated that mTORC2 signaling effectively regulates liver regeneration by inducing oval cell proliferation. Liver progenitor cells or bile duct cells, rather than mature hepatocytes, would be the major source of ductular reaction in DDC-induced cholestatic liver injury.

Automatic speech recognition (ASR) of overlapped speech remains a highly challenging task to date. To this end, multi-channel microphone array data are widely used in state-of-the-art ASR systems. Motivated by the invariance of visual modality to acoustic signal corruption, this paper presents an audio-visual multi-channel overlapped speech recognition system featuring tightly integrated separation front-end and recognition back-end. A series of audio-visual multi-channel speech separation front-end components based on \textit{TF masking}, \textit{filter\&sum} and \textit{mask-based MVDR} beamforming approaches were developed. To reduce the error cost mismatch between the separation and recognition components, they were jointly fine-tuned using the connectionist temporal classification (CTC) loss function, or a multi-task criterion interpolation with scale-invariant signal to noise ratio (Si-SNR) error cost. Experiments suggest that the proposed multi-channel AVSR system outperforms the baseline audio-only ASR system by up to 6.81\% (26.83\% relative) and 22.22\% (56.87\% relative) absolute word error rate (WER) reduction on overlapped speech constructed using either simulation or replaying of the lipreading sentence 2 (LRS2) dataset respectively., Comment: submitted to Interspeech 2020

The reconfigurable intelligent surface (RIS) is considered as a promising new technology for reconfiguring wireless communication environments. To acquire the channel information accurately and efficiently, we only turn on a fraction of all the RIS elements, formulate a sub-sampled RIS channel, and design a deep learning based scheme to extrapolate the full channel information from the partial one. Specifically, inspired by the ordinary differential equation (ODE), we set up connections between different data layers in a convolutional neural network (CNN) and improve its structure. Simulation results are provided to demonstrate that our proposed ODE-based CNN structure can achieve faster convergence speed and better solution than the cascaded CNN.

We present NetReduce, a novel RDMA-compatible in-network reduction architecture to accelerate distributed DNN training. Compared to existing designs, NetReduce maintains a reliable connection between end-hosts in the Ethernet and does not terminate the connection in the network. The advantage of doing so is that we can fully reuse the designs of congestion control and reliability in RoCE. In the meanwhile, we do not need to implement a high-cost network protocol processing stack in the switch, as IB does. The prototype implemented by using FPGA is an out-of-box solution without modifying commodity devices such as NICs or switches. For the coordination between the end-host and the switch, NetReduce customizes the transport protocol only on the first packet in a data message to comply with RoCE v2. The special status monitoring module is designed to reuse the reliability mechanism of RoCE v2 for dealing with packet loss. A message-level credit-based flow control algorithm is also proposed to fully utilize bandwidth and avoid buffer overflow. We study the effects of intra bandwidth on the training performance in multi-machines multi-GPUs scenario and give sufficient conditions for hierarchical NetReduce to outperform other algorithms. We also extend the design from rack-level aggregation to more general spine-leaf topology in the data center. NetReduce accelerates the training up to 1.7x and 1.5x for CNN-based CV and transformer-based NLP tasks, respectively. Simulations on large-scale systems indicate the superior scalability of NetReduce to the state-of-the-art ring all-reduce.

Spintronics, different from traditional electronics, explores additional routes to manipulate electrons through its spin degree of freedom. During its development, magnetic tunnel junctions (MTJs) have been playing a significant role. This is partially due to their fantastic properties such as symmetry-conserving tunneling, which leads to very large magnetoresistance (TMR) and sensitive detection of spin configuration. The relatively simple core structure of the ferromagnet/tunneling barrier/ferromagnet makes it possible to incorporate MTJs into the well-developed fabrication process and a large variety of material systems. Moreover, in recent years, the voltage-controlled effects discovered at FM/Oxide interfaces enables researchers to explore magneto-electric properties of MTJ and more functionalities have been proposed and verified especially in magnetic switching. In this work, we focus our attention on the voltage-controlled effects based on the conventional CoFeB/MgO magnetic tunnel junctions. After a brief introduction to the fundamentals of spin-dependent transport and perpendicular magnetic anisotropy (PMA) in the CoFeB/MgO MTJs. We divert our efforts to various voltage-controlled mechanism discovered in recent years. Chapter 2 briefly touches on the basic experimental technique used in this work. In Chapter 3, we will introduce a revised MTJ nanopillar fabrication procedure which dramatically lowers the requirement for fabricating high-quality MTJ and increases the yield significantly. Chapter 4 will involve interfacial engineering at CoFeB/MgO interface in MTJs. By dusting a thin layer of heavy metal (Ir), in addition to controlling the annealing condition, the VCMA coefficient can be significantly enhanced and the voltage induced switching energy can be greatly reduced, which demonstrated the promises to incorporate MTJs into next-generation memory applications. In Chapter 5, a novel voltage-controlled exchange bias (VCEB) effect is reported and its impl

Lin Tang,1 Meng Xu,2,3 Long Zhang,4 Lin Qu,5 Xiaoyan Liu6 1College of Mathematics and Computer Science, Chifeng University, Chifeng, The Inner Mongol Autonomous Region 024005, People’s Republic of China; 2Department of Urology, The First Affiliated Hospital of Jinzhou Medical University, Jinzhou, Liaoning 121000, People’s Republic of China; 3R&D Department, Seekgene Technology Co., Ltd, Beijing 100000, People’s Republic of China; 4Department of Hepatobiliary Surgery, Yidu Central Hospital, Weifang, Shandong 262500, People’s Republic of China; 5Department of Orthopaedic Surgery, Anshan Hospital of the First Hospital of China Medical University, Anshan, Liaoning 114000, People’s Republic of China; 6Department of Pathology, The Fifth Medical Center of Chinese PLA General Hospital, Beijing 100000, People’s Republic of ChinaCorrespondence: Meng XuJinzhou Medical University, Songpo Road, Jinzhou, Liaoning 121000, People’s Republic of ChinaEmail lnyxyxm@163.comAbstract: In prostate cancer, distant organ metastasis is the leading cause of patient death. Although the mechanism of malignant tumor metastasis is unclear, studies have confirmed that integrin αVβ 3 plays an important role in this process. In prostate cancer, αVβ 3 mediates adhesion, invasion, immune escape and neovascularization through interactions with different ligands. Among these ligands and in addition to proteins that are directly related to tumor invasion, other proteins that contain the RGD structure could also bind to αVβ 3 and cause a number of biological effects. In this article, we summarized the ligand and downstream proteins related to αVβ 3-mediated prostate cancer metastasis as well as some diagnostic and therapeutic measures targeting αVβ 3.Keywords: prostate cancer, metastasis, αVβ 3, signal pathway