Abstract:Recently, many benchmark datasets have been found to contain noisy labels caused by unavoidable human mistakes. Many researchers propose new noise-aware loss functions to achieve robust classification performance. However, we discover that existing noise-aware loss functions cannot fully heal the damage caused by the noise. On the other hand, some methods filter out low confidence samples and train new models, whereas the filtered samples contain both noisy and hard samples that are critical for the robustness of models. Based on the above two discoveries, we devised the Noise-aware Network (NA-Net) for robust training with noisy labels. Each layer of NA-Net contains three groups of convolution kernels responsible for mix samples, clean samples, and noisy samples, termed as mix-kernels, clean-kernels, and noise-kernels, respectively. Mix-kernels are used for finding the clean samples with a newly devised noise-immune (NI) loss function; clean-kernels are targeted at learning better features without being misguided by noise; noise-kernels are trained by the remaining samples to rectify wrong labels for the next iteration. Meanwhile, for increasing the classification performance of mix-kernels, the extracted feature maps of clean-kernels without being poisoned are combined as the input of mix-kernels of the next layer. Also, the knowledge distillation strategy is adopted to distill the knowledge from clean-kernels to the noise-kernels. Extensive experiments demonstrate that the mutual promotion of three groups of kernels in NA-Net achieves state-of-the-art performance on both artificial noisy datasets and real-world datasets.

Novel Neural Networks Applying for Eliminating Noises of Fiber Optic Gyro

Projection Pursuit Learning Networks Applying for Eliminating Noises of Fiber Optic Gyroscope

Noise is the Fatal Poison: A Noise-aware Network for Noisy Dataset Classification

The Study for Optimization of Chromatographic Condition by Means of Artificial Neural Networks

Novel Neural Networks for Eliminating Errors Existing in Fog

Demonstration of photonic neural network for fiber nonlinearity compensation in long-haul transmission systems

A Novel Intelligent Strategy for Improving Measurement Precision of Fog

Depicting Network Structures from Variable Data Produced by Unknown Colored-Noise Driven Dynamics

End-to-End Learning for 100G-PON Based on Noise Adaptation Network

Fiber neural networks for the intelligent optical fiber communications

Optical backpropagation training method and its applications

Artificial Neural Networks for Fiber Nonlinear Noise Estimation

Fiber-Optic Sensor Spectrum Noise Reduction Based on a Generative Adversarial Network

OSNR and nonlinear noise power estimation for optical fiber communication systems using LSTM based deep learning technique

Application of BP Neural Networks in Non-Linearity Correction of Optical Tweezers

Efficient Deep Learning of Nonlinear Fiber-Optic Communications Using a Convolutional Recurrent Neural Network

Optimized Feedforward Neural Network Training for Efficient Brillouin Frequency Shift Retrieval in Fiber.

Efficient On-Chip Learning for Optical Neural Networks Through Power-Aware Sparse Zeroth-Order Optimization

Ultralow complexity long short-term memory network for fiber nonlinearity mitigation in coherent optical communication systems

Research on Orbital Angular Momentum Multiplexing Communication System Based on Neural Network Inversion of Phase

Optical Performance Monitoring Using Lifelong Learning with Confrontational Knowledge Distillation in 7-Core Fiber for Elastic Optical Networks