An official website of the United States government
We propose a data-driven approach for deep convolutional neural network compression that achieves high accuracy with high throughput and low memory requirements. Current network compression methods either find a low-rank factorization of the features that requires more memory, or select only a subset of features by pruning entire filter channels. We propose the Cascaded Projection (CaP) compression method that projects the output and input filter channels of successive layers to a unified low dimensional space based on a low-rank projection. We optimize the projection to minimize classification loss and the difference between the next layer’s features in the compressed and uncompressed networks. To solve this non-convex optimization problem we propose a new optimization method of a proxy matrix using back propagation and Stochastic Gradient Descent (SGD) with geometric constraints. Our cascaded projection approach leads to improvements in all critical areas of network compression: high accuracy, low memory consumption, low parameter count and high processing speed. The proposed CaP method demonstrates state-of-the-art results compressing VGG16 and ResNet networks with over 4× reduction in the number of computations and excellent performance in top-5 accuracy on the ImageNet dataset before and after fine-tuning.
more »
« less
Structural Compression of Convolutional Neural Networks with Applications in Interpretability
Deep convolutional neural networks (CNNs) have been successful in many tasks in machine vision, however, millions of weights in the form of thousands of convolutional filters in CNNs make them difficult for human interpretation or understanding in science. In this article, we introduce a greedy structural compression scheme to obtain smaller and more interpretable CNNs, while achieving close to original accuracy. The compression is based on pruning filters with the least contribution to the classification accuracy or the lowest Classification Accuracy Reduction (CAR) importance index. We demonstrate the interpretability of CAR-compressed CNNs by showing that our algorithm prunes filters with visually redundant functionalities such as color filters. These compressed networks are easier to interpret because they retain the filter diversity of uncompressed networks with an order of magnitude fewer filters. Finally, a variant of CAR is introduced to quantify the importance of each image category to each CNN filter. Specifically, the most and the least important class labels are shown to be meaningful interpretations of each filter.
more »
« less
- Award ID(s):
- 1741340
- PAR ID:
- 10373938
- Date Published:
- Journal Name:
- Frontiers in Big Data
- Volume:
- 4
- ISSN:
- 2624-909X
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Convolutional neural networks (CNNs) are a foundational model architecture utilized to perform a wide variety of visual tasks. On image classification tasks CNNs achieve high performance, however model accuracy degrades quickly when inputs are perturbed by distortions such as additive noise or blurring. This drop in performance partly arises from incorrect detection of local features by convolutional layers. In this work, we develop a neuroscience-inspired unsupervised Sleep Replay Consolidation (SRC) algorithm for improving convolutional filter’s robustness to perturbations. We demonstrate that sleep- based optimization improves the quality of convolutional layers by the selective modification of spatial gradients across filters. We further show that, compared to other approaches such as fine- tuning, a single sleep phase improves robustness across different types of distortions in a data efficient manner.more » « less
-
In recent years, the use of convolutional neural networks (CNNs) for raw resting-state electroencephalography (EEG) analysis has grown increasingly common. However, relative to earlier machine learning and deep learning methods with manually extracted features, CNNs for raw EEG analysis present unique problems for explainability. As such, a growing group of methods have been developed that provide insight into the spectral features learned by CNNs. However, spectral power is not the only important form of information within EEG, and the capacity to understand the roles of specific multispectral waveforms identified by CNNs could be very helpful. In this study, we present a novel model visualization-based approach that adapts the traditional CNN architecture to increase interpretability and combines that inherent interpretability with a systematic evaluation of the modelviaa series of novel explainability methods. Our approach evaluates the importance of spectrally distinct first-layer clusters of filters before examining the contributions of identified waveforms and spectra to cluster importance. We evaluate our approach within the context of automated sleep stage classification and find that, for the most part, our explainability results are highly consistent with clinical guidelines. Our approach is the first to systematically evaluate both waveform and spectral feature importance in CNNs trained on resting-state EEG data.more » « less
-
In this paper, we propose a deep multimodal fusion network to fuse multiple modalities (face, iris, and fingerprint) for person identification. The proposed deep multimodal fusion algorithm consists of multiple streams of modality-specific Convolutional Neural Networks (CNNs), which are jointly optimized at multiple feature abstraction levels. Multiple features are extracted at several different convolutional layers from each modality-specific CNN for joint feature fusion, optimization, and classification. Features extracted at different convolutional layers of a modality-specific CNN represent the input at several different levels of abstract representations. We demonstrate that an efficient multimodal classification can be accomplished with a significant reduction in the number of network parameters by exploiting these multi-level abstract representations extracted from all the modality-specific CNNs. We demonstrate an increase in multimodal person identification performance by utilizing the proposed multi-level feature abstract representations in our multimodal fusion, rather than using only the features from the last layer of each modality-specific CNNs. We show that our deep multi-modal CNNs with multimodal fusion at several different feature level abstraction can significantly outperform the unimodal representation accuracy. We also demonstrate that the joint optimization of all the modality-specific CNNs excels the score and decision level fusions of independently optimized CNNs.more » « less
-
There have been many recent attempts to extend the successes of convolutional neural networks (CNNs) from 2-dimensional (2D) image classification to 3-dimensional (3D) video recognition by exploring 3D CNNs. Considering the emerging growth of mobile or Internet of Things (IoT) market, it is essential to investigate the deployment of 3D CNNs on edge devices. Previous works have implemented standard 3D CNNs (C3D) on hardware platforms, however, they have not exploited model compression for acceleration of inference. This work proposes a hardware-aware pruning approach that can fully adapt to the loop tiling technique of FPGA design and is applied onto a novel 3D network called R(2+1)D. Leveraging the powerful ADMM, the proposed pruning method achieves simultaneous high accuracy and significant acceleration of computation on FPGA. With layer-wise pruning rates up to 10× and negligible accuracy loss, the pruned model is implemented on a Xilinx ZCU102 FPGA board, where the pruned model achieves 2.6× speedup compared with the unpruned version, and 2.3× speedup and 2.3× power efficiency improvement compared with state-of-the-art FPGA implementation of C3D.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.3389/fdata.2021.704182
