In this paper, we focus on comparing the imputation performance of different deep and machine learning techniques on nine related datasets containing different missing rates ranging from 10% to 50%. Moreover, since each feature value ranges differently, such as total liability/equity ratio and earnings per share, the effect of feature normalisation on the imputation results is also examined to see whether normalising the feature values after missing value imputation can improve the prediction model performance. The experimental results show that the deep neural network technique does not necessarily perform better than the traditional machine learning technique for missing value imputation. In particular, the random forest imputation model performs the best, whereas the k-nearest neighbour method is the second best imputation model in terms of the AUC rates and type II errors. The performance in improvement of prediction models after performing feature normalisation is heavily dependent on the chosen classification technique. There is thus no need to consider the normalisation step when the random forest classifier is used. It is found that the deep neural network and support vector machine classifiers can significantly outperform those without feature normalisation.

Abstract This paper designs a novel method to reduce the coding complexity of 3D‐HEVC encoder by utilizing the properties of human visual perception. Two vision‐oriented edge detections are proposed: for colour texture detection, the authors adopt the Just‐Noticeable Distortion (JND); for depth map, the authors combine the Sample Adaptive Offset (SAO) and the Just Noticeable Depth Difference (JNDD) model. The authors also analyse the properties of colour texture and depth map to classify the coding tree unit (CTU) into various kinds of types, including complex‐edge CTU, moderate‐edge CTU and homogeneous CTU. Besides, fast mode decisions and early termination criteria are performed individually on each type of CTUs according to their characteristics. Especially for those CTUs with more edge information, the proposed projection‐based fast mode decision and residual‐based early termination preserve important colour texture while speeding up the coding at the same time. The proposed vision‐oriented algorithm reduces 31.981% of the overall average coding time with only 1.580% BD‐Bitrate increase. Experimental results show that the proposed algorithm can provide considerable time‐saving while still maintain the video quality, which outperforms the previous researches.

A low-cost mobile sensor network structure with a sensing mechanism is presented to explore the border of unknown target areas effectively. At first, mobile sensors with GPS receivers are organized into groups called SensorGroup (SG) which works well for the detecting jobs. After that, an advanced structure called centralized sensor group (CSG) is proposed to further reduce the device cost since only the center node of the group is equipped with a GPS receiver. To detect the target area, an S-shaped sensing mechanism is also presented to cooperate with SG and CSG. The operation processes and detection phases are provided and verified. The detection performance is evaluated by detection completeness and accuracy in different scenarios in a square area of 100 m*100 m sensing environment. The simulation results show that CSG has good detection accuracy and the same performance as that of SG.

The visitor location register (VLR) is used to store temporary records for roaming users. Due to the limited capacity, the VLR may be full. This phenomenon is called VLR overflow. By replacing an existing record, the incoming user can still receive services. The most-idle replacement (MR) policy was proposed to improve the random replacement (RR) policy. The MR policy is almost the optimum method and outperforms RR policy significantly. However, the MR policy requires global information to find the most-idle user. It needs considerable extra cost and is unfeasible due to the separated VLR databases in the real mobile network. Therefore, we propose the second chance replacement (SCR) policy to overcome these drawbacks. The SCR policy needs only an additional bit in the record. Partial information is sufficient for our method, so it is feasible in the real mobile network. Moreover, the simulation results show that the SCR policy outperforms the MR policy in most scenarios investigated in our study.

One of the main challenges in personal communication service (PCS) is to locate many mobiles that may move frequently from place to place. Such a system operation is called location tracking. Many network signals flow, and database queries are required to achieve such a task. In addition to the two-level hierarchical strategy in IS-41 and GSM, several strategies have been proposed to improve the efficiency of location tracking. Pointer forwarding was used to reduce the expensive home location register (HLR) accesses. Previously, the distributed HLR scheme was proposed to prevent the HLR from becoming a bottleneck in the signaling network. However, the length of a forwarding pointer chain may be lengthened in a distributed HLR environment. We propose a more efficient strategy to overcome this potential problem. This strategy attempts to migrate the locating chains in a distributed HLR system when a mobile issues a registration operation. As a consequence, the length of any forwarding pointer chain does not exceed one in our strategy. Simulation results indicate that our strategy significantly decreases the locating cost. In fact, this strategy provides an upper bound of location tracking time owing to the fact that the length of any locating path does not exceed one. Furthermore, obsolete entries in local databases visitor location registers can be reclaimed in this strategy.