An Efficient Stochastic Gradient Descent Algorithm to Maximize the Coverage of Cellular Networks

Liu, Yaxi; Huangfu, Wei; Zhang, Haijun; Long, Keping

doi:10.1109/twc.2019.2914040

Cited by 56 publications

(22 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Adam algorithm [39] is different from traditional stochastic gradient descent [40], [41]. In stochastic gradient descent, a single learning rate is used to update all weights, and the learning rate does not change during the training process.…”

Section: Overall Optimization Of the Rbfnnmentioning

confidence: 99%

Radial Basis Function Neural Network Model for Dissolved Oxygen Concentration Prediction Based on an Enhanced Clustering Algorithm and Adam

Sun

Feng

2021

IEEE Access

View full text Add to dashboard Cite

In fishery aquaculture, water quality directly determines the economic benefits of aquatic products, and dissolved oxygen is an important factor affecting water quality. To accurately grasp the trends of variation in dissolved oxygen, a dissolved oxygen concentration forecasting model based on an enhanced clustering algorithm and Adam with a radial basis function neural network (ECA-Adam-RBFNN) is proposed. An enhanced clustering algorithm (ECA) combining K-means with ant colony optimization is introduced in place of random selection to determine the center positions of the neural network hidden layer units. If the number of center points is too high, the neural network will be overfit, whereas if it is too low, sudden changes will appear in the results. Once the hidden layer centers have been determined, the radial basis function (RBF) width is calculated from the maximum center distance and the number of center points to avoid the two extreme cases of RBF that are too peaked or flat. The recursive least squares (RLS) algorithm is introduced to obtain the connection weights from the hidden layer to the output layer. The Adam algorithm is introduced to iteratively differentiate the objective function to adjust the center values, weights and width while adaptively varying the learning rates for these three types of parameters. Finally, the improved forecasting algorithm is applied for the prediction of the dissolved oxygen concentration in fishery aquaculture. The experimental results show that under identical conditions, compared with a long short-term memory (LSTM) network, a backpropagation neural network (BPNN), a traditional RBF neural network, a support vector regression (SVR) model, an autoregressive integrated moving average (ARIMA), K-MLPNN (K-means muhilayer perceptron neural networks), and SC-K-means-RBF model, the improved algorithm achieves significant reductions in the mean absolute error (MAE), mean absolute percentage error (MAPE) and root mean square error (RMSE) as model evaluation indicators.

show abstract

Section: Overall Optimization Of the Rbfnnmentioning

confidence: 99%

Radial Basis Function Neural Network Model for Dissolved Oxygen Concentration Prediction Based on an Enhanced Clustering Algorithm and Adam

Sun

Feng

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Mean accuracy MIoU FCN with Adam (10) 0.9775 0.9660 0.9403 FCN with AdaGrad (9) 0.9768 0.9654 0.9386 FCN with SGD (7) 0.9782 0.9685 0.9423 DeepLabv2 (5) 0.9713 0.9663 0.9389 Proposed method 0.9824 0.9816 0.9577 voting method, the fuzzy integral is used for the fusion of multiple FCNs with various optimal methods. Experimental results show that FCN-8s has a better performance than FCN-16s and FCN-32s.…”

Section: Accuracy Overall Accuracymentioning

confidence: 99%

“…For a deep learning network, there are some optimal methods for improving the network's performance. The most common method is stochastic gradient descent (SGD), (7) which randomly selects a certain number of training samples in each time for training. This method can usually learn training samples effectively, but it depends on the learning rate setting and the training time is long.…”

Section: Introductionmentioning

confidence: 99%

Using Fully Convolutional Networks for Floor Area Detection

Lin¹,

Li²,

Lee³

2020

Sensors and Materials

View full text Add to dashboard Cite

Most mobile robots use visual images to obtain information about the surrounding environment and the nonlinear diffusion method to detect candidate areas of the floor, but they could not be applied to more complicated environments. In this study, a hybrid of fully convolutional networks (FCNs) and fuzzy integral is proposed for detecting the position of the floor and nonfloor from visual images. FCN is an end-to-end, pixels-to-pixels network for semantic segmentation. Semantic segmentation aims to perform dense segmentation tasks on images and segments each pixel to a specified category. To overcome the majority decision drawback in the traditional voting method and increase the accuracy, the fuzzy integral is used for the fusion of multiple FCNs with various optimal methods. The overall accuracy, mean accuracy, and mean intersection over union (MIoU) of the proposed method are 0.9824, 0.9816, and 0.9577, respectively. The experimental results show that the proposed hybrid method has better accuracy than other methods in identifying the location of the floor area.

show abstract

“…to efficiently explore the solution space in the search for near-optimal solutions. Alternatively, some methods use local search algorithms (e.g., coordinate descent [33], Nelder-Mead [34], gradient descent [14], [35], simulated annealing [36], case-based learning [10], Tabu search [37], primal-dual [38]. .…”

Section: Related Workmentioning

confidence: 99%

A Computationally Efficient Method for QoE-Driven Self-Planning of Antenna Tilts in a LTE Network

2020

View full text Add to dashboard Cite

In future mobile communications systems, network management procedures must be upgraded to consider user quality of experience (QoE) to deal with service diversity. In this work, a computationally efficient centralized method for determining the best configuration of antenna tilts when planning a LTE network is presented. Unlike previous network-centric methods, the proposed self-planning method is driven by QoE criteria. The core of the method is the grouping of cells into clusters without mutual interference, which speeds up the search for the optimal solution with a classical steepest descent algorithm. Method assessment is carried out in a static system-level simulator adjusted with real connection traces. For this purpose, traffic demand in the scenario is broken down per location and service to estimate the QoE obtained by each antenna tilt plan. During the analysis, the method is compared with legacy tilt planning approaches. Results show that the proposed method achieves a near-optimal solution for the overall system QoE with a computational cost lower than state-of-the art algorithms previously reported in the literature.

show abstract

An Efficient Stochastic Gradient Descent Algorithm to Maximize the Coverage of Cellular Networks

Cited by 56 publications

References 31 publications

Radial Basis Function Neural Network Model for Dissolved Oxygen Concentration Prediction Based on an Enhanced Clustering Algorithm and Adam

Radial Basis Function Neural Network Model for Dissolved Oxygen Concentration Prediction Based on an Enhanced Clustering Algorithm and Adam

Using Fully Convolutional Networks for Floor Area Detection

A Computationally Efficient Method for QoE-Driven Self-Planning of Antenna Tilts in a LTE Network

Contact Info

Product

Resources

About