Abstract:In recent years, deep neural networks have achieved exciting results in a variety of tasks, and many fields try to introduce neural network techniques. In mobile edge computing, there are not many attempts that build neural network models in service recommendation or QoS (quality-of-service) prediction. The method proposed in this article is an attempt to employ neural network technique for QoS prediction. Compared to the pure use of QoS records, the exploration for context information in QoS prediction also still needs a lot of efforts. But an increasing number of features are highly likely to result in overfitting problem, especially in the case that the data size is small. To solve those problems, in this article, we propose several new techniques, including denoising auto-encoder with fuzzy clustering (DAFC) and recombination embedding network, focusing on how to use context information and how to alleviate overfitting problem. DAFC uses the denoising auto-encoder, which helps the fuzzy clustering algorithm overcome the defect that the performance is easy to be impacted by the number of clusters. Extensive experiments under different data densities show that these two network structures indeed improve the performance and reduce the overfitting problem.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to use neural network technology for Quality of Service (QoS) prediction in the mobile edge computing environment to recommend appropriate services. Specifically, the paper addresses the following challenges: 1. **QoS Prediction under Sparse Data**: In the mobile edge computing environment, the known service invocation records are usually very sparse, which causes the collaborative - filtering - based methods to perform poorly in similarity calculation and neighbor identification, and further affects the prediction accuracy. 2. **Utilization of Contextual Information**: Most of the existing QoS prediction methods are based on historical invocation records while ignoring the influence of contextual information (such as network connection speed, physical configuration of the running platform, geographical location of the user, etc.) on QoS. These contextual factors can significantly affect the QoS value, but the existing methods lack the ability to effectively utilize this information. 3. **Over - fitting Problem**: As the number of features increases, especially in the case of a small amount of data, the model is prone to the over - fitting problem. The method proposed in the paper aims to alleviate this problem through a new neural network structure. To address the above challenges, the paper proposes the following solutions: - **Combination of Denoising Autoencoder and Fuzzy Clustering (DAFC)**: By combining the denoising autoencoder (DAE) and the fuzzy clustering algorithm (FC), deep - level features are learned and the quality of neighbor identification is improved. DAE helps to eliminate invalid information and reduce the over - fitting problem. - **Reorganized Embedding Network (REN)**: A new neural network structure is designed, which can effectively combine different types of features (such as invocation records and contextual information) and control the reorganization of the embedded feature vectors, further improving the generalization ability of the model. - **Multilayer Perceptron (MLP)**: A multilayer perceptron is used to perform nonlinear feature extraction on the original embedded vectors and the reorganized vectors, and finally generate the prediction results. The paper verifies the effectiveness of the proposed framework and methods through experiments. In particular, under high data sparsity, the new neural network structure can significantly reduce the over - fitting problem and improve the accuracy of QoS prediction.

QoS Prediction for Service Recommendation With Features Learning in Mobile Edge Computing Environment

Recurrent Neural Network Based Collaborative Filtering for QoS Prediction in IoV

A Novel Deep-Learning-Based QoS Prediction Model for Service Recommendation Utilizing Multi-Stage Multi-Scale Feature Fusion With Individual Evaluations

A Deep-Learning Model for Service QoS Prediction Based on Feature Mapping and Inference

Time-Aware and Sparsity-Tolerant QoS Prediction Based on Collaborative Filtering

QoS Prediction of Web Services Based on Two-Phase K-Means Clustering

Location-aware Collaborative Filtering and Feature Interaction Learning for QoS Prediction

Predicting Quality of Services Based on a Two-stream Deep Learning Model with User and Service Graphs

Context-aware QoS Prediction for Web Service Recommendation and Selection.

Learning QoE of Mobile Video Transmission with Deep Neural Network: A Data-Driven Approach

Predicting QoE of Video Streaming with Network-Context Features

Applying a Deep-Learning Approach to Predict the Quality of Web Services

Personalised QoS-based Web Service Recommendation with Service Neighbourhood-Enhanced Matrix Factorisation

Personalized QoS Prediction of Cloud Services via Learning Neighborhood-based Model

Context-Aware and Adaptive QoS Prediction for Mobile Edge Computing Services

A Spatial-Temporal QoS Prediction Approach for Time-aware Web Service Recommendation.

A Self-Adaptive Approach to Service Deployment under Mobile Edge Computing for Autonomous Driving.

Feature Noise Resilient for QoS Prediction with Probabilistic Deep Supervision

Robust QoS Prediction based on Reputation Integrated Graph Convolution Network

Leveraging Federated Learning and Edge Computing for Recommendation Systems within Cloud Computing Networks

No-Reference Qoe Prediction Model for Video Streaming Service in 3g Networks