Model compression for faster structural separation of macromolecules captured by Cellular Electron Cryo-Tomography

@article{Guo2018ModelCF,
  title={Model compression for faster structural separation of macromolecules captured by Cellular Electron Cryo-Tomography},
  author={JiaLiang Guo and Bo Zhou and Xiangrui Zeng and Zachary Freyberg and Min Xu},
  journal={Image Analysis and Recognition: International Conference, ICIAR ... : proceedings. ICIAR},
  year={2018},
  volume={10882},
  pages={
          144-152
        }
}
  • JiaLiang GuoBo Zhou Min Xu
  • Published 31 January 2018
  • Computer Science
  • Image Analysis and Recognition: International Conference, ICIAR ... : proceedings. ICIAR
Electron Cryo-Tomography (ECT) enables 3D visualization of macromolecule structure inside single cells. Macromolecule classification approaches based on convolutional neural networks (CNN) were developed to separate millions of macromolecules captured from ECT systematically. However, given the fast accumulation of ECT data, it will soon become necessary to use CNN models to efficiently and accurately separate substantially more macromolecules at the prediction stage, which requires additional… 

Semi-supervised Macromolecule Structural Classification in Cellular Electron Cryo-Tomograms using 3D Autoencoding Classifier

This work proposes a 3D autoencoding classifier framework for simultaneous macromolecule structural reconstruction and classification and demonstrates the effectiveness of this approach against other semi-supervised classification approaches on both real and simulated datasets.

One-Shot Learning With Attention-Guided Segmentation in Cryo-Electron Tomography

This paper develops a one- shot learning framework, called cryo-ET one-shot network (COS-Net), for simultaneous classification of macromolecular structure and generation of the voxel-level 3D segmentation, using only one training sample per class.

Template-Based and Template-Free Approaches in Cellular Cryo-Electron Tomography Structural Pattern Mining

This chapter presents three major Cryo-ET structural pattern mining approaches to give an overview of traditional methods and recent advances in Crye-ET data analysis.

Open-set Recognition of Unseen Macromolecules in Cellular Electron Cryo-Tomograms by Soft Large Margin Centralized Cosine Loss

A novel loss function for deep neural networks to extract discriminative features for unseen macromolecular structure recognition in CECT, called Soft Large Margin Centralized Cosine Loss (Soft LMCCL), which is a useful design in the detection task of unseen structures and is potentially useful in other similar open-set scenarios.

Feature Decomposition Based Saliency Detection in Electron Cryo-Tomograms

A saliency detection method that computes the likelihood that a subregion in a tomogram stands out from the background, and can remove the majority of the background region, and significantly speed up the subsequent processing of segmentation and recognition of cellular components captured by ECT.

Adversarial domain adaptation for cross data source macromolecule in situ structural classification in cellular electron cryo-tomograms

A deep learning-based adversarial domain adaptation (3D-ADA) method is adapted to timely address the domain shift problem in CECT data analysis and can improve cross data source recovery of novel macromolecular structures.

A survey of deep learning-based methods for cryo-electron tomography data analysis

Visualize and Compress Single Logo Recognition Neural Network

DeconvNet is used to visualize the network’s feature maps and a metric is proposed to analyze them quantitatively to obtain a better understanding of the influences in the network brought by accompanying datasets.

Limited Angle Tomography Reconstruction: Synthetic Reconstruction via Unsupervised Sinogram Adaptation

An end-to-end limited angle tomography reconstruction adversarial network (Tomo-GAN) via unsupervised sinogram adaptation without having real sinogram-reconstruction pairs is proposed, demonstrating a consistent improvement over previous methods while significantly reducing the reconstruction computation time.

References

SHOWING 1-8 OF 8 REFERENCES

Deep learning-based subdivision approach for large scale macromolecules structure recovery from electron cryo tomograms

This article proposes a new approach for subdividing subtomograms into smaller but relatively homogeneous subsets that is based on supervised structural feature extraction using deep learning, in combination with unsupervised clustering and reference‐free classification.

Improved deep learning-based macromolecules structure classification from electron cryo-tomograms

This work proposed a deep learning-based image classification approach for large-scale systematic macromolecular structure separation from CECT data, and proposes three newly designed individual CNN models: an extended version of (Deep Small Receptive Field) DSRF3D, a 3D residual block-based neural network, and a convolutional 3D (C3D)-based model, CB3D.

New views of cells in 3D: an introduction to electron tomography.

A Survey of Model Compression and Acceleration for Deep Neural Networks

This paper survey the recent advanced techniques for compacting and accelerating CNNs model developed, roughly categorized into four schemes: parameter pruning and sharing, low-rank factorization, transferred/compact convolutional filters, and knowledge distillation.

Deep Compression: Compressing Deep Neural Network with Pruning, Trained Quantization and Huffman Coding

This work introduces "deep compression", a three stage pipeline: pruning, trained quantization and Huffman coding, that work together to reduce the storage requirement of neural networks by 35x to 49x without affecting their accuracy.

Two-Bit Networks for Deep Learning on Resource-Constrained Embedded Devices

This work presents Two-Bit Networks (TBNs) for model compression of CNNs with edge weights constrained to (-2, -1, 1, 2), which can be encoded with two bits, representing a reasonable tradeoff between model size and performance.

Distilling the Knowledge in a Neural Network

This work shows that it can significantly improve the acoustic model of a heavily used commercial system by distilling the knowledge in an ensemble of models into a single model and introduces a new type of ensemble composed of one or more full models and many specialist models which learn to distinguish fine-grained classes that the full models confuse.

Model compression

This work presents a method for "compressing" large, complex ensembles into smaller, faster models, usually without significant loss in performance.