Rate Schedule Research Articles

Aerial Scene Classification through Fine-Tuning with Adaptive Learning Rates and Label Smoothing

Remote Sensing (RS) image classification has recently attracted great attention for its application in different tasks, including environmental monitoring, battlefield surveillance, and geospatial object detection. The best practices for these tasks often involve transfer learning from pre-trained Convolutional Neural Networks (CNNs). A common approach in the literature is employing CNNs for feature extraction, and subsequently train classifiers exploiting such features. In this paper, we propose the adoption of transfer learning by fine-tuning pre-trained CNNs for end-to-end aerial image classification. Our approach performs feature extraction from the fine-tuned neural networks and remote sensing image classification with a Support Vector Machine (SVM) model with linear and Radial Basis Function (RBF) kernels. To tune the learning rate hyperparameter, we employ a linear decay learning rate scheduler as well as cyclical learning rates. Moreover, in order to mitigate the overfitting problem of pre-trained models, we apply label smoothing regularization. For the fine-tuning and feature extraction process, we adopt the Inception-v3 and Xception inception-based CNNs, as well the residual-based networks ResNet50 and DenseNet121. We present extensive experiments on two real-world remote sensing image datasets: AID and NWPU-RESISC45. The results show that the proposed method exhibits classification accuracy of up to 98%, outperforming other state-of-the-art methods.

Suitability of pavement type for developing countries from an economic perspective using life cycle cost analysis

Road infrastructure plays a significant role in enriching economy of a developing country. Sustainability of transportation infrastructure aids in this development process, hence road authorities thrive to build roads with longer service life. Bangladesh being a developing country, road maintenance and management task have become a challenge due to tropical weather condition and traffic stream with overloaded vehicles. Flexible pavements constitute most of the pavements in this country, and it has been found that these pavements lose their serviceability at the end of fifth year instead of sustaining for the design period of twenty years. Under such circumstances, rigid pavement can be a possible alternative. In this study, life cycle cost analysis of both rigid and flexible pavements have been performed for a particular case study of Sylhet-Bholaganj Road Project. Future costs have been estimated using growth factor, then all associated costs have been converted to Net Present Value (NPV) to estimate the life cycle cost. The life cycle cost analysis was performed using the Roads and Highways Department (RHD) Schedule of Rates 2015 and 2002 version for performing sensitivity analysis and it has revealed that rigid pavement is economically superior to flexible pavement. The need of frequent periodic maintenance of flexible pavement causes its life cycle cost to increase significantly whereas rigid pavements do not need costly periodic maintenance. Rigid pavements follow the build and forget concept which eventually makes it economically viable and sustainable type of pavement especially for developing countries facing overloading phenomenon under tropical climate.

Abstract 2100: Deep learning can predict microsatellite instability from histology in colorectal cancer across different ethnic groups

Abstract Microsatellite instability (MSI) is known to be an important genomic marker for prediction of a patient's response to immune checkpoint inhibitors across various types of cancer. Conventional methods to measure DNA mismatch repair (MMR) status are pCR-based panel sequencing or immunohistochemical staining based methods. Recently, deep learning has been proposed as an efficient diagnostic tool in this regard, as it can learn to predict MSI directly from H&E stained slides. However, it is thought that deep learning may not generalize effectively beyond the ethnicities present in the training dataset. Therefore, we investigated the impact of ethnic differences on deep learning performance at MSI classification across a range of model architectures and hyperparameters. A total of 360 WSIs from a subset of the TCGA colorectal cancer (TCGA set) and another 170 WSIs from Gangnam Severance Hospital (Yonsei University College of Medicine, Seoul, Republic of Korea, Gangnam set) were included in this study. 360 H&E stained WSIs from TCGA set were divided into a set of square tiles of size 256x256um (0.5um/pixel) and then color normalized using Macenko normalization. Those tiles were divided into two classes - MSI-H and MSS (Microsatellite stable) based on the patient's MSI status. We trained various models on those tiles using a range of hyperparameters including model architecture, number of trainable layers, learning rate, and learning rate scheduling. The performance of this model was measured using the AUROC in the TCGA set (internal validation) and Gangnam set (external validation). Our results for ResNet18 - a popular convolutional network architecture - on TCGA and Gangnam test datasets are around 0.79 and 0.76 respectively, with a standard deviation on Gangnam of 0.03. Given good hyperparameters and model architecture, the model's performance on the Gangnam test data was within one standard deviation of its performance on the TCGA test data, despite the fact that said model was trained using only the TCGA data containing few asian patients. The lack of a significant performance difference challenges the notion that there are morphological differences in colorectal cancers between patients of Caucasian and Asian ethnicities that would prevent a deep learning system from generalizing effectively. Citation Format: Isaiah S. Pressman, Hongming Xu, Jeonghyun Kang, Yoon Jin Cha, Sung Hak Lee, Tae Hyun Hwang. Deep learning can predict microsatellite instability from histology in colorectal cancer across different ethnic groups [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2100.

Loan guarantees and credit supply

The efficiency of federal lending guarantees depends on whether guarantees increase lending supply or simply act as a subsidy to lenders. We use notches in the guarantee rate schedule for Small Business Administration (SBA) loans to estimate the elasticity of bank lending volume to loan guarantees. We show significant bunching in the loan distribution on the side of the size threshold that carries a more generous loan guarantee. The excess mass implies that increasing guarantee generosity by one percentage point of loan principal would increase per-loan lending volume by $19,000. Excess mass increases in periods with guarantee generosity, and placebo results indicate that the effect disappears when the guarantee notch is eliminated.

Controlled workplace charging of electric vehicles: The impact of rate schedules on transformer aging

To accelerate adoption of non-residential charging for electric vehicles, sites must maximize utilization of existing electrical infrastructure. In this study we model electric vehicle charging at a workplace using real charging data and evaluate the lifetime of the site’s transformer as the number of charging stations is incrementally increased. We implement and compare a range of control schemes for workplace charging including minimizing the peak load, capping the total load, minimizing bills under different rate structures with time-of-use energy costs and demand charges, and directly minimizing the transformer’s aging. These are compared by the number of vehicles they allow the transformer to support, the transformer’s health, and the operator’s electricity bill. We draw a connection between minimizing the peak load and improving the transformer’s health. We observe that minimizing the electricity bill is the best scheme by both criteria when the bill includes a demand charge; in our experiment it allowed the infrastructure to support over 67% more cars than under uncontrolled charging. To protect the transformer we recommend that demand charges or capacity management be applied to parking lots of charging electric vehicles with high infrastructure utilization, and operators schedule charging to minimize their electricity bills.

Lung Sound Classification Using Snapshot Ensemble of Convolutional Neural Networks.

We propose a robust and efficient lung sound classification system using a snapshot ensemble of convolutional neural networks (CNNs). A robust CNN architecture is used to extract high-level features from log mel spectrograms. The CNN architecture is trained on a cosine cycle learning rate schedule. Capturing the best model of each training cycle allows to obtain multiple models settled on various local optima from cycle to cycle at the cost of training a single mode. Therefore, the snapshot ensemble boosts performance of the proposed system while keeping the drawback of expensive training of ensembles moderate. To deal with the class-imbalance of the dataset, temporal stretching and vocal tract length perturbation (VTLP) for data augmentation and the focal loss objective are used. Empirically, our system outperforms state-of-the-art systems for the prediction task of four classes (normal, crackles, wheezes, and both crackles and wheezes) and two classes (normal and abnormal (i.e. crackles, wheezes, and both crackles and wheezes)) and achieves 78.4% and 83.7% ICBHI specific micro-averaged accuracy, respectively. The average accuracy is repeated on ten random splittings of 80% training and 20% testing data using the ICBHI 2017 dataset of respiratory cycles.

Joint source coding rate allocation and flow scheduling for data aggregation in collaborative sensing networks

In collaborative sensing networks such as WSNs (Wireless Sensor Networks), due to overlapping converge areas among neighbor nodes, they may percept a large number of similar or identical data, which incurs sensing data redundancy and unnecessary energy consumption. Slepian-wolf theorem based source coding is an effective method to reduce redundancy in data aggregation. However, the exponential growth of constraints prevent the method from practical application. Therefore, in this paper, we propose a cross-layer optimization framework to solve data aggregation problem by jointly considering optimal source encoding rate and flow scheduling. By proving the convex of constraints of Spelian-Wolf theorem, we relax original constraints so that the optimal encoding rate scheme can be adopted. The relaxation makes the optimization problem feasible. Furthermore, we employ dual decomposition to separate the original problem into two sub-problems. By solving the two subproblem distributedly, we provide optimal encoding rate allocation for each node and flow scheduling for each link. Simulation results demonstrate that our framework can reduce data redundancy and network traffic significantly compared to the exiting algorithms.

A Novel Learning Rate Schedule in Optimization for Neural Networks and It’s Convergence

The process of machine learning is to find parameters that minimize the cost function constructed by learning the data. This is called optimization and the parameters at that time are called the optimal parameters in neural networks. In the process of finding the optimization, there were attempts to solve the symmetric optimization or initialize the parameters symmetrically. Furthermore, in order to obtain the optimal parameters, the existing methods have used methods in which the learning rate is decreased over the iteration time or is changed according to a certain ratio. These methods are a monotonically decreasing method at a constant rate according to the iteration time. Our idea is to make the learning rate changeable unlike the monotonically decreasing method. We introduce a method to find the optimal parameters which adaptively changes the learning rate according to the value of the cost function. Therefore, when the cost function is optimized, the learning is complete and the optimal parameters are obtained. This paper proves that the method ensures convergence to the optimal parameters. This means that our method achieves a minimum of the cost function (or effective learning). Numerical experiments demonstrate that learning is good effective when using the proposed learning rate schedule in various situations.

Rapid-Infusion Rituximab in a Pediatric Population.

The use of rapid rituximab infusion in certain pediatric populations has generally been regarded as safe. The safety of our institution's rapid rituximab protocol was evaluated. The primary end point was the number of and severity of adverse drug reactions. Secondary end points included a description of the patient population defined by the indication, dose, and number of rituximab infusions administered. Additionally, the difference in infusion times in hours of those receiving rapid rituximab infusions versus the theoretical infusion time of subsequent administration rate schedules was defined. A total of 88 infusions for 22 patients were reviewed. No dose-limiting adverse reactions were observed. Three patients experienced grade 1 isolated infusion-related adverse events during a single infusion encounter. Two of the three patients received additional doses of rapid rituximab infusions without incident, whereas the other patient no longer required rituximab therapy. The use of a 90-minute rituximab infusion protocol in pediatric patients with non-rheumatic diseases was well tolerated.

Event Study Analysis of Political and Economical Factors Influencing Bank Notes Exchange Rate Dynamics in the First Third of the 19th c.

The article studies Russian ruble bank notes dynamics in the first half of the 19th century. In particular, it addresses political and economical factors for this dynamics and empirically evaluates the influence of important military, political and economical events on bank notes exchange rate. The author analyzes the direction and the intensity of this influence during the period which is up to three months beginning with the initial start of these factors.  To form time series of the bank notes exchange rate from 1800 to 1839 the article examines Moskovskie Vedomosti that published Moscow stock exchange rates schedules as well as Sankt-Peterburzhskie Vedomosti and Kommercheskaya Gazeta which published Saint Petersburg stock exchange rates schedules. Time series analysis of daily exchange rates has shown that they were more sensitive to endogenous and economical events rather than exogenous and political ones. The study has found out that an exchange rate is more sensitive to internal political and economical reasons rather than external political and political ones. For instance, the sensitivity to cholera was stronger than that to wars. One sees that stock exchange demonstrated more sensitivity to events hindering foreign trade and uncertainty as far as foreign policy is concerned. In addition, it has been found out that Moscow exchange rate was much more sensitive than Saint Petersburg one. However, the later demonstrated longer sensitivity to events. .

Protective efficacy of the Newcastle disease virus genotype VII–matched vaccine in commercial layers

Newcastle disease virus (NDV) is a major threat to the poultry industry worldwide, with a diversity of genotypes associated with severe economic losses in all poultry sectors. Class II genotype VII NDV are predominant in the Middle East and Asia, despite intensive vaccination programs using conventional live and inactivated NDV vaccines. In Egypt, the disease is continuously spreading, causing severe economical losses in the poultry industry. In this study; the protective efficacy of a commercial, inactivated recombinant genotype VII NDV–matched vaccine (KBNP-C4152R2L strain) against challenge with the velogenic NDV strain (Chicken/USC/Egypt/2015) was evaluated in commercial layers. Two vaccination regimes were used; live NDV genotype II (LaSota) vaccine on days 10, 18, and 120, with either the inactivated NDV genotype II regime or inactivated NDV genotype VII–matched vaccine regime on days 14, 42, and 120. The 2 regimes were challenged at the peak of egg production on week 26. Protection by the 2 regimes was evaluated after experimental infection, based on mortality rate, clinical signs, gross lesions, virus shedding, seroconversion, and egg production schedule. The results show that these 2 vaccination regimes protected commercial layer chickens against mortality, but some birds showed mild clinical signs and reduced egg production temporarily. However, the combination of live NDV genotype II and recombinant inactivated genotype VII vaccines provided better protection against virus shedding (20% and 0% vs. 60% and 40%) as assessed in tracheal swabs and (20% and 0% vs. 20% and 20%) in cloacal swabs collected at 3 and 5 D post challenge (dpc), respectively. In addition, egg production levels in birds receiving the inactivated NDV genotype VII–matched vaccine regime and in those given inactivated genotype II vaccines were 76.6, 79, 82, and 87.4% and 77.7, 72.5, 69, and 82.5% at 7, 14, 21, and 28 dpc, respectively. The results of this study indicate that recombinant genotype-matched inactivated vaccine along with a live attenuated vaccine can reduce virus shedding and improve egg production in commercial layers challenged with a velogenic genotype VII virus under field conditions. This regime may ensure a proper control strategy in layers.

SWOC analysis of Mahatma Gandhi National Rural Employment Guarantee Scheme (MGNREGS) in Dharmapuri District of Tamil Nadu

This study was conducted in the Pennagaram block of Dharmapuri district with 120 respondents for identifying the strengths, weakness, opportunities and challenges of MGNREGS. The major strengths are the availability of work with assured wages during the lean season, increase in the bargaining power of agricultural labor in rural areas, increasing their purchasing power of agricultural laborers and assist in proper repair and maintenance of community assets. The weakness identified are delay in payments and defects in carrying out the schedule of rates, lack of planning and expertise in identifying development work and prioritizing them in village panchayats, lack of awareness about social audit and e- muster rolls. With regard to opportunities, it has ensured women’s participation in the labour force leading to their empowerment, ensures food and nutrient security, creates job opportunities for all stakeholders in their rural habitats and prevents and checks migration to nearby towns/cities in search of employment. The challenges are poor allocation of funds and its disbursal to the beneficiaries affecting its functioning, Poor MGNREGS implementation and monitoring resulting in the creation of less public assets and infrastructure facilities, wide spread corruption and misuse of allocated funds and increased political interference in the selection of beneficiaries and assigning of works.

Brachytherapy in head and neck malignancies: Indian Brachytherapy Society (IBS) recommendations and guidelines.

PurposeBrachytherapy (BT) forms major treatment modality in squamous cell carcinoma of head and neck cancers (HNC). However, there is a dearth of literature and guidelines for the use in various indications. High-dose-rate brachytherapy (HDR-BT) in Indian scenario is an important treatment modality, and the recommendations in this guidelines aim to provide the necessary recommendations for the use of HDR-BT for uniform application across the country in patients with HNC.Material and methodsA panel consisting of members of the Indian Brachytherapy Society (IBS), based on their clinical experience was invited. The process involved defining important steps, precautions, target volumes and indications, thorough literature review, and discussion with fellow members. The guidelines were established and formulated the recommendations for HDR-BT based on available evidences and individual experience for sites, relevant to Indian settings.ResultsThe IBS recommends the use of HDR brachytherapy as a part of treatment of head and neck tumors. The scope of these guidelines and recommendations included practical suggestions, ensuring efficient use of brachytherapy treatment as radical with external beam radiotherapy (EBRT) boost, palliative and adjuvant as definitive, or re-radiation as salvage for HNC in India. The IBS has made specific site-wise recommendations for previously untreated and recurrent HNC patients on their selection criteria, implant techniques, target volume definition, and HDR treatment parameters, such as time, dose rate, total dose, and fractionation schedules. Limited experience exists with HDR-BT in patients with head and neck cancers in India and across the globe.ConclusionsIBS provided a consensus statement and guidelines for the head and neck brachytherapy and believed that these recommendations will overcome the fear of practicing radiation oncologists. This should generate interest amongst students and will help radiation oncologists all across the country to use the art of brachytherapy carefully in HNC patients, with better curative and salvage options.

A Constitutional Wealth Tax

Policymakers and scholars are giving serious consideration to a federal wealth tax. Wealth taxation could address the harms from rising economic inequality, promote equality of social and economic opportunity, and raise the revenue needed to fund critical government programs. These reasons for taxing wealth may not matter, however, if a federal wealth tax is unconstitutional. Scholars debate whether a tax on a wealth base (a “traditional wealth tax”) would be a “direct tax” subject to apportionment among the states by population. This Article argues, in contrast, that this possible constitutional restriction on a traditional wealth tax may not matter. If the Court struck down a traditional wealth tax, Congress could instead tax wealth by adjusting a taxpayer’s income tax liability on account of her wealth. This Article describes three general methods for making this adjustment (collectively, “Wealth Integration” methods). A taxpayer’s wealth could affect her taxable income base (the “Base Method”), the applicable rate schedule (the “Rate Method”), or the availability of credits against tax (the “Credit Method”). Wealth Integration methods could replicate the economic effects of a traditional wealth tax but with an intrinsically different constitutional analysis. The Court could strike down Wealth Integration methods only by overruling settled prior precedent, invalidating many current features of the income tax, and fundamentally restricting Congress’s power to tax income under the Sixteenth Amendment. Finally, the possibility that Congress could instead tax wealth through Wealth Integration methods provides a new argument why the Court should uphold a traditional wealth tax. Otherwise, the Court would have to choose between fundamentally restricting the Sixteenth Amendment—and jeopardizing the income tax as we know it—or distinguishing between economically similar taxes on the basis of their formal label while still allowing Congress to tax wealth and diminishing the effect of the apportionment requirement as a restraint on Congress’s taxing power.

PACL: Piecewise Arc Cotangent Decay Learning Rate for Deep Neural Network Training

Deep neural networks (DNNs) are currently the best-performing method for many classification problems. For training DNNs, the learning rate is the most important hyper-parameter, choice of which affects the performance of the model greatly. In recent years, some learning rate schedulers, such as HTD, CLR, and SGDR, have been proposed. These methods, some of which make use of the cycling mechanism to improve the convergence speed and accuracy of DNN, but performance degradation occurs in the convergence process. Others have good accuracy, but their convergence speed is too slow. This paper proposed a new learning rate schedule called piecewise arc cotangent decay learning rate (PACL), which can not only improve the convergence speed and accuracy of DNN but also significantly reduce performance degradation zone caused by the cycling mechanism. It is easy to implement, but almost at no extra computing expense. Finally, we demonstrate the effectiveness of PACL, on training CIFAR-10, CIFAR-100, and Tiny ImageNet with ResNet, DenseNet, WRN, SEResNet, and MobileNet.

Learning Rate Scheduling Policies

With the availability of high processing capability hardwares at less expensive prices, it is possible to successfully train multi-layered neural networks. Since then, several training algorithms have been developed, from algorithms which are statically initialized to algorithms which adaptively change. It is observed that to improve the training process of neural networks, the hyper-parameters are to be fine tuned. Learning Rate, Decay rate, number of epochs, number of hidden layers and number of neurons in the network are some of the hyper-parameters in concern. Of these, the Learning rate plays a crucial role in enhancing the learning capability of the network. Learning rate is the value by which the weights are adjusted in a neural network with respect to the gradient descending towards the expected optimum value. This paper discusses four types of learning rate scheduling which helps to find the best learning rates in less number of epochs. Following these scheduling methods, facilitates to find better initial learning rate value and step-wise updation during the later phase of the training process. In addition the discussed learning rate schedules are demonstrated using COIL-100, Caltech-101 and CIFAR-10 datasets trained on ResNet. The performance is evaluated using the metrics, Precision, Recall and F1-Score. The results analysis show that, depending on the nature of the dataset, the performance of the Learning Rate Scheduling policy varies. Hence the choice of the scheduling policy to train a neural network is made, based on the data.

Integer linear programming formulations for the variable data rate and variable channel bandwidth scheduling problem in wireless networks

The IEEE 802.11ac standard enables a higher transmission speed than the previous IEEE standards because it allows the network frequency spectrum to be divided into communication channels of different bandwidths, varying from 20MHz to 160MHz. In this paper, we introduce the Variable Rate and Variable Bandwidth Scheduling Problem (VRBSP), which is a generalization of the classical Variable Rate Scheduling Problem (VRSP) on wireless networks. The best algorithm in the literature of VRSP, Datarate PPTAS, can only be efficiently run on networks with up to 64 links and does not guarantee optimality. In this paper, we propose two Mixed Integer Linear Programming (MILP) formulations that are used within a MILP solver to seek optimal schedules for VRBSP. This approach can also be used to solve VRSP. The computational experiments were carried out on classical network instances from the literature with up to 2048 links. They show that the MILP-based exact algorithms were able to find optimal VRBSP schedules for networks with up to 256 links and optimal VRSP schedules for networks with up to 1024 links within 3600 seconds of running time.

Forecast-Informed Energy Storage Utilization in Local Area Power Systems

Distributed energy storage is beneficial to energy consumers and system operators, but it is expensive to install and requires careful management. To improve the viability of demand-side energy storage, this study presents a new method of obtaining optimal control actions for a demand-side energy storage system in the presence of distributed generation. The optimal usage policy is determined by framing the selection of charge and discharge actions as a stochastic optimization problem, which is conveniently represented as a Markov decision process. The proposed method uses short-term forecasts of loads and local generation to represent the time-dependence and nonstationarity of the net load profile. In contrast to previous studies, the forecast information is used to restrict the problem state space, reducing the computational complexity of the policy calculation. The method is tested using historic load and generation data with real time-of-use rate schedules and achieves substantial reductions in energy costs over similar existing methods.

Why hasn't Social Security changed since 1977?

Despite increasing earnings inequality and aging population, there has been no major reform of Social Security since 1977. To account for this puzzling observation, I construct a heterogeneous agent model with incomplete financial markets. There is a government who chooses the replacement rate schedule by maximizing the weighted welfare of all generations alive. I specify the Pareto weights as a function of household's age and identify them both in 1977 and 2018. I find that there was a slight bias towards younger workers in 1977. However, the bias has shifted towards older households in 2018. This shift is a result of counteracting forces induced by: 1) population aging and 2) increased labor productivity risk and increased college premium. The effect of population aging turns out to be stronger quantitatively.

A Research on QoS Optimization in 4G Cellular Networks

cellular system conveys sight and sound website online traffic productively during ongoing transmission. in view that this device transmits blended media information, there may be a prerequisite to adjust the management price as indicated via the gadget situations an excellent manner to hold up fruitful device hobby. in the end, there is a need to allot the assets capably. Nature of provider (QoS) improvement within the sort of system is additionally trying out specially in 4G cellular systems. severa plans were proposed for QoS development in cellular structures which be a part of name Admission control (CAC) and electricity control Scheme. Be that as it is able to, CAC with electricity control isn't always sufficient to offer the most ideal asset use. alongside the ones lines, it is straightforward to build up an powerful QoS primarily based definitely resource Allocation and Scheduling Scheme for normal sight and sound site visitors. The primary objective of proposed studies paintings is to build up a nice of provider based resource Allocation and Adaptive rate Scheduling (QRAAS) plot in cellular systems to improve prepare dependability, boom the framework limit and enhance framework proficiency and therefore to improve best of service in cell systems. To execute the formerly stated QRAAS conspire, resource Allocation based definitely call Admission manage (RACAC) plan and Adaptive price Scheduling (ARS) plot were created.