结合注意力的3D卷积网络脑胶质瘤分割算法

为了提升脑胶质瘤分割精度，提出一种结合注意力机制的3D卷积神经网络算法。输入3个不同尺度的图像块，经过9个卷积层和1个分类层后得到3个不同的分类结果，将分类结果与注意力学习到的权重相乘并逐体素相加得到输出。此外该算法采用了一种混合Dice损失函数与Focal损失函数的超参数损失函数。实验表明，该算法的Dice系数在整体区域、核心区域以及增强区域分别达到了95.31%、80.12%、82.25%。与已有的一种脑胶质瘤分割算法deepmedic相比，整体区域、核心区域以及增强区域的Dice系数分别提升了3%、2%、6%。在脑胶质瘤分割方面，具有重要的临床意义。     

大数据时代大学生个人隐私保护问题研究

伴随着互联网+的风生水起,信息量暴增的大数据时代应运而来,"网络原住民"大学生不可避免、理所当然地处于大数据洪流之中,但也为其个人隐私保护带来了极大挑战。文章针对大数据带来的安全隐私问题,指出大学生个人隐私泄露的可能原因,提出可以采取的个人隐私保护措施,给出大数据时代隐私保护几种技术,期望能给相关人士提供有效参考。     

Local reverse entropy weighted LBF model solving by Split Bregman for image segmentation

Multimedia Tools and Applications - In this paper, an efficient level set model is proposed for image segmentation. Firstly, the original local binary fitting (LBF) model is redefined as a weighted...     

An experimental study of relative total variation and probabilistic collaborative representation for iris recognition

Multimedia Tools and Applications - Iris images collected under different conditions often suffer from specular reflections, cast shadows, motion blur, defocus blur, occlusion caused by eyelashes...     

低活化马氏体钢热变形行为及机理型本构建模研究

利用Gleeble-3500热模拟试验机研究了低活化马氏体钢在变形温度为850～950 ℃、应变速率为0.001～1 s^-1条件下的热变形行为。建立了流变应力本构方程，并评估了该方程的预测能力。绘制了低活化马氏体钢在不同应变下的热加工图。结果表明：在较高的应变速率条件下，该材料主要发生动态回复，在较高变形温度和较低应变速率下具有明显的动态再结晶特征；本构方程的预测结果与实验结果符合良好；变形温度870～930 ℃、应变速率0.001～0.01 s^-1和变形温度920～950 ℃、应变速率0.3～1 s^-1分别是真应变为0.4和0.6下最优的热加工区域。     

Cross-linked polyvinyl amidoxime fiber: a highly selective and recyclable adsorbent of gallium from Bayer liquor

Gallium, a scarce metal produced mainly from Bayer liquor, is widely used in semiconductors. Ion-exchange method is currently considered as the most effective method to recover gallium from Bayer liquor. In this article, fibrous amidoxime adsorbents are introduced to recover gallium from Bayer liquor. In addition, hydrazine cross-linked polyvinyl amidoxime (HPAO) fibers have been prepared. The structure of the as-prepared fibers was ascertained by FTIR, elemental analysis and weight gain rate. The adsorption kinetics, adsorption isotherm and recycling performance were investigated by batch method. Cross-linking duration was studied and it turned out to be an important factor to optimize the adsorption capacity and recycling performance. After 1.5 h cross-linking time, the fiber showed the highest gallium adsorption capacity of 14.83 mg/g in Bayer liquor. After 3 h cross-linking time, the fiber showed the best recycling performance, which retained 82.9% of the initial adsorption ability after four cycles. Adsorption capacity of vanadium was lower than 1 mg/g for all samples. HPAO displayed very fast adsorption kinetics with an equilibrium at 60 min. The EDS results confirmed the low extraction of aluminum and vanadium provided by HPAO fibers. The gallium-loaded fiber could be effectively eluted by acidified thiourea. With proper control of the cross-linking duration by hydrazine, HPAO fiber with high selectivity towards gallium, high adsorption capacity and good recycle performance could be obtained, which is promising for recovering gallium needed for industrial applications.     

全局力平衡PDC钻头布齿优化设计

针对聚晶金刚石复合片(PDC)钻头在钻井破岩过程中因受力不平衡，导致钻井倾斜、井径扩大、钻头产生横向振动和涡动，造成钻头早期失效的问题，基于已有PDC钻头切削力学知识，构建了PDC钻头弯曲力矩及全局力平衡布齿优化设计模型，并提出模型求解方法。布齿设计实例结果表明，采用全局力平衡布齿设计方法得到的布齿结构可使PDC钻头在不同进尺条件下均能保持全局力平衡状态，尤其低进尺（小于1 mm）时的力平衡状态得到极大改善，提高了PDC钻头的钻井稳定性，对于改善钻头受力、提升钻头使用寿命、提高钻井质量与效率具有重要意义。     

基于并行多通道卷积长短时记忆网络的轴承寿命预测方法

在预测轴承剩余使用寿命时，数据间的时序特性是一个可以利用的重要隐藏信息。为了更好地提取具有时序信息的特征用于预测，提出了一种基于并行多通道卷积长短时记忆网络（PMCCNN-LSTM）的剩余使用寿命预测模型。该模型主要由两部分组成：前端为并行多通道卷积网络（PMCCNN），提取信号特征，挖掘数据的时序特性，并采用逐层训练和微调的方式提升参数的收敛性；后端为长短时记忆（LSTM）网络，基于特征进行剩余使用寿命预测，并采用加权平均的方法对预测结果进行平滑处理。在一个轴承加速寿命实验的公开数据集上使用留一法验证了该模型的准确性，实验结果表明：所提模型的平均误差与最大误差分别比传统的卷积神经网络（CNN）低23.38%和15.84%，比传统的LSTM低24.14%和19.01%，比卷积长短时记忆网络（CNN-LSTM）低30.32%和23.09%。