In recent years, deep neural networks have been successful in both industry and academia, especially for computer vision tasks. The great success of deep learning is mainly due to its scalability to encode large-scale data and to maneuver billions of model parameters. However, it is a challenge to deploy these cumbersome deep models on devices with limited resources, e.g., mobile phones and embedded devices, not only because of the high computational complexity but also the large storage requirements. To this end, a variety of model compression and acceleration techniques have been developed. As a representative type of model compression and acceleration, knowledge distillation effectively learns a small student model from a large teacher model. It has received rapid increasing attention from the community. This paper provides a comprehensive survey of knowledge distillation from the perspectives of knowledge categories, training schemes, teacher-student architecture, distillation algorithms, performance comparison and applications. Furthermore, challenges in knowledge distillation are briefly reviewed and comments on future research are discussed and forwarded.

Knowledge Distillation: A Survey

Deep neural models, in recent years, have been successful in almost every field. However, these models are huge, demanding heavy computation power. Besides, the performance boost is highly dependent on redundant labeled data. To achieve faster speeds and to handle the problems caused by the lack of labeled data, knowledge distillation (KD) has been proposed to transfer information learned from one model to another. KD is often characterized by the so-called ‘Student-Teacher’ (S-T) learning framework and has been broadly applied in model compression and knowledge transfer. This paper is about KD and S-T learning, which are being actively studied in recent years. First, we aim to provide explanations of what KD is and how/why it works. Then, we provide a comprehensive survey on the recent progress of KD methods together with S-T frameworks typically used for vision tasks. In general, we investigate some fundamental questions that have been driving this research area and thoroughly generalize the research progress and technical details. Additionally, we systematically analyze the research status of KD in vision applications. Finally, we discuss the potentials and open challenges of existing methods and prospect the future directions of KD and S-T learning.

Knowledge Distillation and Student-Teacher Learning for Visual Intelligence: A Review and New Outlooks.

In recent years, deep neural networks have been successful in both industry and academia, especially for computer vision tasks. The great success of deep learning is mainly due to its scalability to encode large-scale data and to maneuver billions of model parameters. However, it is a challenge to deploy these cumbersome deep models on devices with limited resources, e.g., mobile phones and embedded devices, not only because of the high computational complexity but also the large storage requirements. To this end, a variety of model compression and acceleration techniques have been developed. As a representative type of model compression and acceleration, knowledge distillation effectively learns a small student model from a large teacher model. It has received rapid increasing attention from the community. This paper provides a comprehensive survey of knowledge distillation from the perspectives of knowledge categories, training schemes, teacher–student architecture, distillation algorithms, performance comparison and applications. Furthermore, challenges in knowledge distillation are briefly reviewed and comments on future research are discussed and forwarded.

Deep neural models, in recent years, have been successful in almost every field, even solving the most complex problem statements. However, these models are huge in size with millions (and even billions) of parameters, demanding heavy computation power and failing to be deployed on edge devices. Besides, the performance boost is highly dependent on redundant labeled data. To achieve faster speeds and to handle the problems caused by the lack of labeled data, knowledge distillation (KD) has been proposed to transfer information learned from one model to another. KD is often characterized by the so-called 'Student-Teacher' (S-T) learning framework and has been broadly applied in model compression and knowledge transfer. This paper is about KD and S-T learning, which are being actively studied in recent years. First, we aim to provide explanations of what KD is and how/why it works. Then, we provide a comprehensive survey on the recent progress of KD methods together with S-T frameworks typically used for vision tasks. In general, we investigate some fundamental questions that have been driving this research area and thoroughly generalize the research progress and technical details. Additionally, we systematically analyze the research status of KD in vision applications. Finally, we discuss the potentials and open challenges of existing methods and prospect the future directions of KD and S-T learning. 

/pdf/knowledge-distillation-and-student-teacher-learning-for-1udp0atw.pdf

Knowledge Distillation and Student-Teacher Learning for Visual Intelligence: A Review and New Outlooks

Machine remaining useful life (RUL) prediction is vital in improving the reliability of industrial systems and reducing maintenance cost Recently, long short-term memory (LSTM) based algorithms have achieved state-of-the-art performance for RUL prediction due to their strong capability of modeling sequential sensory data In many cases, the RUL prediction algorithms are required to be deployed on edge devices to support real-time decision making, reduce the data communication cost, and preserve the data privacy However, the powerful LSTM-based methods which have high complexity cannot be deployed to edge devices with limited computational power and memory To solve this problem, we propose a knowledge distillation framework, entitled KDnet-RUL, to compress a complex LSTM-based method for RUL prediction Specifically, it includes a generative adversarial network based knowledge distillation (GAN-KD) for disparate architecture knowledge transfer, a learning-during-teaching based knowledge distillation (LDT-KD) for identical architecture knowledge transfer, and a sequential distillation upon LDT-KD for complicated datasets We leverage simple and complicated datasets to verify the effectiveness of the proposed KDnet-RUL The results demonstrate that the proposed method significantly outperforms state-of-the-art KD methods The compressed model with 128 times less weights and 462 times less total float point operations even achieves a comparable performance with the complex LSTM model for RUL prediction

KDnet-RUL: A Knowledge Distillation Framework to Compress Deep Neural Networks for Machine Remaining Useful Life Prediction

The audio classification task aims to discriminate between different audio signal types. In this task, deep neural networks have achieved better performance than the traditional shallow architecture-based machine-learning method. However, deep neural networks often require huge computational and storage requirements that hinder the deployment in embedded devices. In this paper, we proposed a distillation method which transfers knowledge from well-trained networks to a small network, and the method can compress model size while improving audio classification precision. The contributions of the proposed method are two folds: a multi-level feature distillation method was proposed and an adversarial learning strategy was employed to improve the knowledge transfer. The extensive experiments are conducted on three audio classification tasks, audio scene classification, general audio tagging, and speech command recognition. The experimental results demonstrate that: the small network can provide better performance while achieves the calculated amount of floating-point operations per second (FLOPS) compression ratio of 76:1 and parameters compression ratio of 3:1.

/pdf/an-adversarial-feature-distillation-method-for-audio-2082r74eho.pdf

An Adversarial Feature Distillation Method for Audio Classification

Recently, deep learning has achieved state-of-the-art performance in more aspects than traditional shallow architecture-based machine-learning methods. However, in order to achieve higher accuracy, it is usually necessary to extend the network depth or ensemble the results of different neural networks. Increasing network depth or ensembling different networks increases the demand for memory resources and computing resources. This leads to difficulties in deploying depth-learning models in resource-constrained scenarios such as drones, mobile phones, and autonomous driving. Improving network performance without expanding the network scale has become a hot topic for research. In this paper, we propose a cross-architecture online-distillation approach to solve this problem by transmitting supplementary information on different networks. We use the ensemble method to aggregate networks of different structures, thus forming better teachers than traditional distillation methods. In addition, discontinuous distillation with progressively enhanced constraints is used to replace fixed distillation in order to reduce loss of information diversity in the distillation process. Our training method improves the distillation effect and achieves strong network-performance improvement. We used some popular models to validate the results. On the CIFAR100 dataset, AlexNet's accuracy was improved by 5.94%, VGG by 2.88%, ResNet by 5.07%, and DenseNet by 1.28%. Extensive experiments were conducted to demonstrate the effectiveness of the proposed method. On the CIFAR10, CIFAR100, and ImageNet datasets, we observed significant improvements over traditional knowledge distillation.

https://www.mdpi.com/1099-4300/21/4/357/pdf

Multistructure-Based Collaborative Online Distillation

. The Transformer-based encoder-decoder architecture has re-cently made significant advances in recognizing handwritten mathematical expressions. However, the transformer model still suffers from the lack of coverage problem, making its expression recognition rate (ExpRate) inferior to its RNN counterpart. Coverage information, which records the alignment information of the past steps, has proven effective in the RNN models. In this paper, we propose CoMER, a model that adopts the coverage information in the transformer decoder. Specifically, we propose a novel Attention Refinement Module (ARM) to refine the attention weights with past alignment information without hurting its parallelism. Furthermore, we take coverage information to the extreme by proposing self-coverage and cross-coverage, which utilize the past alignment information from the current and previous layers. Experiments show that CoMER improves the ExpRate by 0.61%/2.09%/1.59% compared to the current state-of-the-art model, and reaches 59.33%/59.81%/62.97% on the CROHME 2014/2016/2019 test sets. 1

CoMER: Modeling Coverage for Transformer-based Handwritten Mathematical Expression Recognition

Surface defect detection is one of the most essential processes for industrial quality inspection. Deep learning-based surface defect detection methods have shown great potential. However, the well-performed models usually require large training data and can only detect defects that appeared in the training stage. When facing incremental few-shot data, defect detection models inevitably suffer from catastrophic forgetting and misclassification problem. To solve these problems, this paper proposes a new knowledge distillation network, called Dual Knowledge Align Network (DKAN). The proposed DKAN method follows a pretraining-finetuning transfer learning paradigm and a knowledge distillation framework is designed for fine-tuning. Specifically, an Incremental RCNN is proposed to achieve decoupled stable feature representation of different categories. Under this framework, a Feature Knowledge Align (FKA) loss is designed between class-agnostic feature maps to deal with catastrophic forgetting problems, and a Logit Knowledge Align (LKA) loss is deployed between logit distributions to tackle misclassification problems. Experiments have been conducted on the incremental Few-shot NEU-DET dataset and results show that DKAN outperforms other methods on various few-shot scenes, up to 6.65% on the mean Average Precision metric, which proves the effectiveness of the proposed method.

A New Knowledge Distillation Network for Incremental Few-Shot Surface Defect Detection

Aiming at the problems that the target detection accuracy of inland watercraft is reduced, and the difficulty of labeling samples is increased due to fog, according to the characteristics of inland water ship detection, an adversarial unsupervised domain adaptive method is proposed, using labeled sunny samples and unlabeled foggy samples, which can adaptively detect ship targets in foggy conditions. On the basis of Faster-RCNN, a ship local feature domain discriminator based on full convolution and a ship global feature domain discriminator based on a hybrid attention mechanism are respectively constructed to realize the features of sunny and foggy ship images in multi-level domain space alignment, to better complete the adaptive detection from sunny ship data to foggy ship data. At the same time, using the idea of adversarial learning, a local feature adversarial loss function based on the least square method and a global feature adversarial loss based on cross entropy are designed to strengthen the adversarial training between feature extractor and domain discriminator. The experimental results on the inland ship data set show that only using the marked sunny ship data and the unmarked foggy ship data can adaptively detect ships in foggy weather, and the accuracy and real-time performance are significantly higher than those of the comparison model.

Liang Gao

Papers

An Adversarial Feature Distillation Method for Audio Classification

Multistructure-Based Collaborative Online Distillation

CoMER: Modeling Coverage for Transformer-based Handwritten Mathematical Expression Recognition

A New Knowledge Distillation Network for Incremental Few-Shot Surface Defect Detection

Adversarial unsupervised domain adaptive inland vessel detection method