Instance Enhancement Batch Normalization: An Adaptive Regulator of Batch Noise

Batch Normalization (BN) (Ioffe and Szegedy 2015) normalizes the features of an input image via statistics of a batch of images and hence BN will bring the noise to the gradient of training loss. Previous works indicate that the noise is important for the optimization and generalization of deep neural networks, but too much noise will harm the performance of networks. In our paper, we offer a new point of view that the self-attention mechanism can help to regulate the noise by enhancing instance-specific information to obtain a better regularization effect. Therefore, we propose an attention-based BN called Instance Enhancement Batch Normalization (IEBN) that recalibrates the information of each channel by a simple linear transformation. IEBN has a good capacity of regulating the batch noise and stabilizing network training to improve generalization even in the presence of two kinds of noise attacks during training. Finally, IEBN outperforms BN with only a light parameter increment in image classification tasks under different network structures and benchmark datasets.

CC Path Generation Method Based on Cutting Shape of Flat-End Cutter

The calculating algorithm for Cutter Contacting (CC) paths is proposed in this article to enhance machining precision of complex surfaces with a flat-end cutter. With the style of coordinate transformation, effective cutting shape and equavellent machining radius of a flat-end cutter had been calculated and deduced, and the step-over distance formed by two adjacent CC paths were also obtained. The corresponding NURBS path of tool center points including knot vector and controlling points was obtained from the cutter contacting points.The parameter increment along the CC path interval direction was calculated by using geometric analysis of the local area of CC point.

Cutter Contacting Machining Paths Generation Method for Surface Flat-End Milling

In order to enhance machining precision of complex surfaces with a flat-end cutter, the generating method of cutter contacting (CC) paths is presented in this article. With the method of coordinate transformation, effective machining radius of a flat-end cutter had been deduced and the step-over formed by two adjacent CC paths were implemented. The parameter increment along the CC path interval direction was calculated by using geometric analysis of the local area of CC point. The corresponding NURBS path of tool center points including knot vector and controlling points was obtained from the cutter contacting points.

CC Path Generation Method in 5-Axis Surface Machining with a Flat-End Cutter

A 5-axis machining method of Non-Uniform Rational B-Spline (NURBS) surface has been proposed in this paper. With the local curvature and the machining error, the step-size was calculated to get the parameter of the next Cutter-Contacting (CC) point. With the method of coordinate transformation, effective machining radius of a flat-end cutter had been deduced and the step-over formed by two adjacent CC paths were implemented. The parameter increment along the CC path interval direction was calculated by using geometric analysis of the local area of CC point. Finally, once all of CC points on the NURBS surface had been obtained, the CC paths for 5-axis surface milling were generated.