ID 原文 译文
1693 首先，利用 py-torch 深度学习框架建立一个关系型深度学习网络，并使用源域数据对网络进行预训练; Firstly, a relational deep learning network is established by using the pytorch framework, and the network is pre-trained by the source domain data.
1694 然后，使用此网络对目标域数据进行分类预测，将分类概率最大的类标签作为数据的伪标签; Then, the network is used to predict the target domain data, and the la-bel with the highest classification probability is used as the data's pseudo label.
1695 最后，利用目标域的伪标签数据和源域的真实标签数据对网络进行混合训练，并重复伪标签标记与混合训练过程。 Finally, the network is hybrid trained using the pseudo label data of the target domain and the real label data of the source domain, then repeating the pseudo-labeled and hybrid trained process.
1696 实验结果表明，相对于现有主流少样本学习算法，FSLSS模型有更好的泛化能力及知识迁移效果。 The experimental results show that the FSLSS model has better generalization ability and knowledge transfer effect than the existing few-shot learning algorithms.
1697 最小割问题(minimum cut problem)是 NP(Non-deterministic Polynomial)难问题，警示传播算法(warningpropagation)是一种基于因子图的消息传递算法，可用于求解组合优化问题。 The minimum cut problem (MCP)is an NP (Non-deterministic Polynomial)-hard problem, warningpropagation (WP)is a kind of message passing algorithm based on factor graph, it solve the combinatorial optimization problem.
1698 首先，本文借助隐马尔可夫模型将无向图转换为因子图，将求解最小割映射为求解因子图的相应问题。 First, HMM (Hidden Markov Model)converted undirected graph to factor graph.
1699 进而设计一种求解最小割的警示传播算法。 Then, we designed a kind ofwarning propagation algorithm to solving the minimum cut.
1700 最后，选取了几组随机无向图实例进行数值实验，实验结果表明，该算法在求解速度上优于同类算法。 Finally, we selected skit randomly undirected graphs numericalexperiments. The experimental show that the algorithm precedes similar algorithms in speed.
1701 提出了一种非对称双栅应变硅 HALO 掺杂沟道金属氧化物半导体场效应管结构。 The asymmetrical double-material-gate s-Si(strained Silicon)HALO doping channel MOSFET(Metal-Ox-ide-Semiconductor Field Effect Transistor)structure is proposed.
1702 该器件前栅和背栅由两种不同功函数的金属构成，沟道为应变硅 HALO 掺杂沟道，靠近源区为低掺杂区域，靠近漏区为高掺杂区域。 The front gate and back gate are composed of two metalswith different work functions. It has the higher doping concentration in the HALO doping channel end near the drain.