Interstitial Lung Disease （ILD） segmentation labels are highly costly， leading to small sample sizes in existing datasets and resulting in poor performance of trained models. To address this issue， a segmentation algorithm for ILD based on multi-task learning was proposed. Firstly， a multi-task segmentation model was constructed based on U-Net. Then， the generated lung segmentation labels were used as auxiliary task labels for multi-task learning. Finally， a method of dynamically weighting the multi-task loss functions was used to balance the losses of the primary task and the secondary task. Experimental results on a self-built ILD dataset show that the Dice Similarity Coefficient （DSC） of the multi-task segmentation model reaches 82.61%， which is 2.26 percentage points higher than that of U-Net. The experimental results demonstrate that the proposed algorithm can improve the segmentation performance of ILD and can assist clinical doctors in ILD diagnosis.

Most classification models in deep learning-based encrypted traffic classification methods have deep and straight structure with the problem of vanishing gradient， and the increase of the number of network layers leads to significant increase of model structure and computational complexity. Based on these， an encrypted traffic classification method based on improved Inception-ResNet was proposed. In the method， the classification model was constructed by improving the Inception module and embedding it into the convolutional neural network as a residual block in a residual structural connection way. In addition， the loss function of the classification model was improved， and the effectiveness of the proposed method was verified by using VPN-nonVPN dataset. Experimental results show that the proposed method achieves the precision， recall， and F1 score of more than 94.21%， 92.53%， and 93.31%， respectively， in the classification experiments of two senerios. In the comparison experiments with other methods， taking the 12-class classification experiment， which is the most difficult one， as an example， the proposed method is higher than C4.5 decision tree algorithm and 1D-CNN （1 Dimensional-Convolutional Neural Network） by 13.91 and 9.50 percentage points higher in precision and by 14.87 and 1.59 percentage points in recall. Compared with the algorithms such as CAE （Convolutional Auto Encoding） and SAE （Stacked Auto Encoder）， the proposed method not has obvious improvement on the indicators， but has significant shorter single training time， fully demonstrating that the proposed method is a state-of-the-art method.

Concerning the power complementary limitation due to the time-reversed assumption of prototype filter in the design of traditional DFT (Discrete Fourier Transform) modulated filter banks, a time domain modified method was introduced to design the DFT filter banks from the time domain perfect reconstruction perspectives in this paper. Moreover, the designed filter banks were applied to the filter banks based multi-carrier transceivers. The time domain modified method relaxed the time-reversed assumption of prototype filter, that is, the filter banks at the receiver were conjugate transpose form of the filter banks at the transmitter. Moreover, it adopted the time domain formula of the perfect reconstruction property as the solution to design the filter banks at the receiver, which would ensure the perfect reconstruction of filter banks and avoid the power complementary limitation in the design of prototype filter at the same time. Compared to the traditional design method, the time domain modified method improves the design freedom of prototype in the filter banks, so suitable prototype filters could be obtained according to the various application environments without considering power complementary restrictions. Moreover, the time domain modified DFT filter banks based multi-carrier transceivers has a better SER (Symbol Error Ratio) performance in QPSK (Quadrature Reference Phase Shift Keying) modulation, ideal and the 3GPP TS 25.104 pedestrian multipath channel and one-tap frequency-domain equalization.

As a new service for data storage and management, cloud storage has the virtue of portability and simplicity in use. However, it also prompts a significant problem of ensuring the integrity and recovery of data. A data recovery system for cloud storage based on fountain code was designed to resolve the problem. In this system, the user encoded his data by fountain code to make the tampered data recoverable, and tested the data's integrity with Hash functions so that the complexity in data verification and recovery was reduced. Through this system, the user can verify whether his data have been tampered or not by sending a challenge to the servers. Furthermore, once some data have been found tampered, the user can require and supervise the servers to locate and repair them timely. The experimental results show that the data integrity detection precision reaches 99% when the data's manipulation rate is 1%-5%.