This paper proposes a new learning paradigm called filter grafting, which aims to improve the representation capability of Deep Neural Networks (DNNs). The motivation is that DNNs have unimportant (invalid) filters (e.g., l1 norm close to 0). These filters limit the potential of DNNs since they are identified as having little effect on the network. While filter pruning removes these invalid filters for efficiency consideration, filter grafting re-activates them from an accuracy boosting perspective. The activation is processed by grafting external information (weights) into invalid filters. To better perform the grafting process, we develop an entropy-based criterion to measure the information of filters and an adaptive weighting strategy for balancing the grafted information among networks. After the grafting operation, the network has very few invalid filters compared with its untouched state, enpowering the model with more representation capacity. We also perform extensive experiments on the classification and recognition tasks to show the superiority of our method. For example, the grafted MobileNetV2 outperforms the non-grafted MobileNetV2 by about 7 percent on CIFAR-100 dataset. Code is available at https://github.com/fxmeng/filter-grafting.git.
Recent salient object detection (SOD) methods based on deep neural network have achieved remarkable performance. However, most of existing SOD models designed for low-resolution input perform poorly on high-resolution images due to the contradiction between the sampling depth and the receptive field size. Aiming at resolving this contradiction, we propose a novel one-stage framework called Pyramid Grafting Network (PGNet), using transformer and CNN backbone to extract features from different resolution images independently and then graft the features from transformer branch to CNN branch. An a…
Zhejiang University（浙江大学）；Stevens Institute of Technology（史蒂文斯理工学院）；Alibaba-Zhejiang University Joint Research Institute of Frontier Technologies（阿里巴巴-浙江大学前沿技术联合研究院）
Knowledge distillation has demonstrated encouraging performances in deep model compression. Most existing approaches, however, require massive labeled data to accomplish the knowledge transfer, making the model compression a cumbersome and costly process. In this paper, we investigate the practical few-shot knowledge distillation scenario, where we assume only a few samples without human annotations are available for each category. To this end, we introduce a principled dual-stage distillation scheme tailored for few-shot data. In the first step, we graft the student blocks one by one onto the…
Grafting Transformer on Automatically Designed Convolutional Neural Network for Hyperspectral Image Classification
Northwestern Polytechnical University（西北工业大学）；Harbin Institute of Technology, Shenzhen（哈尔滨工业大学（深圳））；Shaanxi Normal University（陕西师范大学）；Yan’an University（延安大学）
Current deep domain adaptation methods used in computer vision have mainly focused on learning discriminative and domain-invariant features across different domains. In this paper, we present a novel "deep adversarial transition learning" (DATL) framework that bridges the domain gap by projecting the source and target domains into intermediate, transitional spaces through the employment of adjustable, cross-grafted generative network stacks and effective adversarial learning between transitions. Specifically, we construct variational auto-encoders (VAE) for the two domains, and form bidirectio…
University of Otago（奥塔哥大学）；Institute of Information Engineering Chinese Academy of Sciences（中国科学院信息工程研究所）
Current deep domain adaptation methods used in computer vision have mainly focused on learning discriminative and domain-invariant features across different domains. In this paper, we present a novel approach that bridges the domain gap by projecting the source and target domains into a common association space through an unsupervised ``cross-grafted representation stacking'' (CGRS) mechanism. Specifically, we construct variational auto-encoders (VAE) for the two domains, and form bidirectional associations by cross-grafting the VAEs' decoder stacks. Furthermore, generative adversarial network…
Deep Convolutional Neural Networks (CNN) have been successfully applied to many real-life problems. However, the huge memory cost of deep CNN models poses a great challenge of deploying them on memory-constrained devices (e.g., mobile phones). One popular way to reduce the memory cost of deep CNN model is to train binary CNN where the weights in convolution filters are either 1 or -1 and therefore each weight can be efficiently stored using a single bit. However, the compression ratio of existing binary CNN models is upper bounded by around 32. To address this limitation, we propose a novel me…
Since the convolutional neural networks are often trained with redundant parameters, it is possible to reduce redundant kernels or filters to obtain a compact network without dropping the classification accuracy. In this paper, we propose a filter pruning method using the hierarchical group sparse regularization. It is shown in our previous work that the hierarchical group sparse regularization is effective in obtaining sparse networks in which filters connected to unnecessary channels are automatically close to zero. After training the convolutional neural network with the hierarchical group …
Zhejiang University（浙江大学）；Fabu Inc.（法布公司）；Alibaba-Zhejiang University Joint Institute of Frontier Technologies（阿里巴巴-浙江大学前沿技术联合研究院）；GAC R&D Center（广汽研究院）
Neural network compression empowers the effective yet unwieldy deep convolutional neural networks (CNN) to be deployed in resource-constrained scenarios. Most state-of-the-art approaches prune the model in filter-level according to the "importance" of filters. Despite their success, we notice they suffer from at least two of the following problems: 1) The redundancy among filters is not considered because the importance is evaluated independently. 2) Cross-layer filter comparison is unachievable since the importance is defined locally within each layer. Consequently, we must manually specify l…
Moscow Institute of Physics and Technology（莫斯科物理技术学院）
The article describes a system for image recognition using deep convolutional neural networks. Modified network architecture is proposed that focuses on improving convergence and reducing training complexity. The filters in the first layer of the network are constrained to fit the Gabor function. The parameters of Gabor functions are learnable and are updated by standard backpropagation techniques. The system was implemented on Python, tested on several datasets and outperformed the common convolutional networks.
Digital artists often improve the aesthetic quality of digital photographs through manual retouching. Beyond global adjustments, professional image editing programs provide local adjustment tools operating on specific parts of an image. Options include parametric (graduated, radial filters) and unconstrained brush tools. These highly expressive tools enable a diverse set of local image enhancements. However, their use can be time consuming, and requires artistic capability. State-of-the-art automated image enhancement approaches typically focus on learning pixel-level or global enhancements. T…
Explicit encoding of group actions in deep features makes it possible for convolutional neural networks (CNNs) to handle global deformations of images, which is critical to success in many vision tasks. This paper proposes to decompose the convolutional filters over joint steerable bases across the space and the group geometry simultaneously, namely a rotation-equivariant CNN with decomposed convolutional filters (RotDCF). This decomposition facilitates computing the joint convolution, which is proved to be necessary for the group equivariance. It significantly reduces the model size and compu…
Filters in a Convolutional Neural Network (CNN) contain model parameters learned from enormous amounts of data. In this paper, we suggest to decompose convolutional filters in CNN as a truncated expansion with pre-fixed bases, namely the Decomposed Convolutional Filters network (DCFNet), where the expansion coefficients remain learned from data. Such a structure not only reduces the number of trainable parameters and computation, but also imposes filter regularity by bases truncation. Through extensive experiments, we consistently observe that DCFNet maintains accuracy for image classification…
Friedrich-Alexander Universität Erlangen-Nürnberg (FAU)（埃尔朗根-纽伦堡弗里德里希-亚历山大大学）；Siemens Healthcare GmbH（西门子医疗有限公司）
Deep neural networks have shown great success in low dose CT denoising. However, most of these deep neural networks have several hundred thousand trainable parameters. This, combined with the inherent non-linearity of the neural network, makes the deep neural network diffcult to understand with low accountability. In this study we introduce JBFnet, a neural network for low dose CT denoising. The architecture of JBFnet implements iterative bilateral filtering. The filter functions of the Joint Bilateral Filter (JBF) are learned via shallow convolutional networks. The guidance image is estimated…
We present a filter correlation based model compression approach for deep convolutional neural networks. Our approach iteratively identifies pairs of filters with the largest pairwise correlations and drops one of the filters from each such pair. However, instead of discarding one of the filters from each such pair naïvely, the model is re-optimized to make the filters in these pairs maximally correlated, so that discarding one of the filters from the pair results in minimal information loss. Moreover, after discarding the filters in each round, we further finetune the model to recover from th…
It is a challenging task to deploy computationally and memory intensive State-of-the-art deep neural networks (DNNs) on embedded systems with limited hardware resources and power budgets. Recently developed techniques like Deep Compression make it possible to fit large DNNs, such as AlexNet and VGGNet, fully in on-chip SRAM. But sparse networks compressed using existing encoding formats, like CSR or CSC, complex the computation at runtime due to their irregular memory access characteristics. In [1], we introduce a computation dataflow, stacked filters stationary dataflow (SFS), and a correspon…
Faculty of Computer and Information Science, University of Ljubljana（卢布尔雅那大学计算机与信息科学学院）；CN-CR Centre, School of Computer Science, University of Birmingham（伯明翰大学计算机学院CN-CR中心）
Classical deep convolutional networks increase receptive field size by either gradual resolution reduction or application of hand-crafted dilated convolutions to prevent increase in the number of parameters. In this paper we propose a novel displaced aggregation unit (DAU) that does not require hand-crafting. In contrast to classical filters with units (pixels) placed on a fixed regular grid, the displacement of the DAUs are learned, which enables filters to spatially-adapt their receptive field to a given problem. We extensively demonstrate the strength of DAUs on a classification and semanti…
We propose a novel particle filter for convolutional-correlation visual trackers. Our method uses correlation response maps to estimate likelihood distributions and employs these likelihoods as proposal densities to sample particles. Likelihood distributions are more reliable than proposal densities based on target transition distributions because correlation response maps provide additional information regarding the target's location. Additionally, our particle filter searches for multiple modes in the likelihood distribution, which improves performance in target occlusion scenarios while dec…
Marquette University（马凯特大学）；University of Florida（佛罗里达大学）
This work proposes a novel framework for visual tracking based on the integration of an iterative particle filter, a deep convolutional neural network, and a correlation filter. The iterative particle filter enables the particles to correct themselves and converge to the correct target position. We employ a novel strategy to assess the likelihood of the particles after the iterations by applying K-means clustering. Our approach ensures a consistent support for the posterior distribution. Thus, we do not need to perform resampling at every video frame, improving the utilization of prior distrib…
Lancaster University（兰卡斯特大学）；University of Lincoln（林肯大学）
We propose a novel online multi-target visual tracker based on the recently developed Hypothesized and Independent Stochastic Population (HISP) filter. The HISP filter combines advantages of traditional tracking approaches like MHT and point-process-based approaches like PHD filter, and it has linear complexity while maintaining track identities. We apply this filter for tracking multiple targets in video sequences acquired under varying environmental conditions and targets density using a tracking-by-detection approach. We also adopt deep CNN appearance representation by training a verificati…
Noise reduction is one the most important and still active research topic in low-level image processing due to its high impact on object detection and scene understanding for computer vision systems. Recently, we can observe a substantial increase of interest in the application of deep learning algorithms in many computer vision problems due to its impressive capability of automatic feature extraction and classification. These methods have been also successfully applied in image denoising, significantly improving the performance, but most of the proposed approaches were designed for Gaussian n…
This paper addresses the problem of fixed motion and measurement models for multi-target filtering using an adaptive learning framework. This is performed by defining target tuples with random finite set terminology and utilisation of recurrent neural networks with a long short-term memory architecture. A novel data association algorithm compatible with the predicted tracklet tuples is proposed, enabling the update of occluded targets, in addition to assigning birth, survival and death of targets. The algorithm is evaluated over a commonly used filtering simulation scenario, with highly promis…
Normalize Filters! Classical Wisdom for Deep Vision
Electrical Engineering and Computer Sciences, University of California, Berkeley（加州大学伯克利分校电子工程与计算机科学系）；Computer Science and Engineering, University of Michigan（密歇根大学计算机科学与工程系）
Modelling and analysis of the 8 filters from the "master key filters hypothesis" for depthwise-separable deep networks in relation to idealized receptive fields based on scale-space theory
KTH Royal Institute of Technology（皇家理工学院）；TU Wien（维也纳技术大学）
Recently, significant progress has been achieved in deep image matting. Most of the classical image matting methods are time-consuming and require an ideal trimap which is difficult to attain in practice. A high efficient image matting method based on a weakly annotated mask is in demand for mobile applications. In this paper, we propose a novel method based on Deep Learning and Guided Filter, called Inductive Guided Filter, which can tackle the real-time general image matting task on mobile devices. We design a lightweight hourglass network to parameterize the original Guided Filter method th…
In this paper, we present substantial evidence that a deep neural network will intrinsically learn the appropriate way to discretize the ideal continuous reconstruction filter. Currently, the Ram-Lak filter or heuristic filters which impose different noise assumptions are used for filtered back-projection. All of these, however, inhibit a fully data-driven reconstruction deep learning approach. In addition, the heuristic filters are not chosen in an optimal sense. To tackle this issue, we propose a formulation to directly learn the reconstruction filter. The filter is initialized with the idea…
Hong Kong Metropolitan University（香港都会大学）；Nanjing University of Aeronautics and Astronautics（南京航空航天大学）；Lingnan University（岭南大学）；Education University of Hong Kong（香港教育大学）
Image filters are fast, lightweight and effective, which make these conventional wisdoms preferable as basic tools in vision tasks. In practical scenarios, users have to tweak parameters multiple times to obtain satisfied results. This inconvenience heavily discounts the efficiency and user experience. We propose basis composition learning on single image filters to automatically determine their optimal formulas. The feasibility is based on a two-step strategy: first, we build a set of filtered basis (FB) consisting of approximations under selected parameter configurations; second, a dual-bran…
Southern University of Science and Technology（南方科技大学）；University of Technology Sydney（悉尼科技大学）；Fudan University（复旦大学）
This paper proposed a Soft Filter Pruning (SFP) method to accelerate the inference procedure of deep Convolutional Neural Networks (CNNs). Specifically, the proposed SFP enables the pruned filters to be updated when training the model after pruning. SFP has two advantages over previous works: (1) Larger model capacity. Updating previously pruned filters provides our approach with larger optimization space than fixing the filters to zero. Therefore, the network trained by our method has a larger model capacity to learn from the training data. (2) Less dependence on the pre-trained model. Large …
Institute of Computing Technology, Chinese Academy of Sciences（中国科学院计算技术研究所）；Faculty of Information Technology, Beijing University of Technology（北京工业大学信息技术学院）
Deep learning has achieved impressive results in many areas, but the deployment of edge intelligent devices is still very slow. To solve this problem, we propose a novel compression and acceleration method based on data distribution characteristics for deep neural networks, namely Pruning Filter via Gaussian Distribution Feature (PFGDF). Compared with previous advanced pruning methods, PFGDF compresses the model by filters with insignificance in distribution, regardless of the contribution and sensitivity information of the convolution filter. PFGDF is significantly different from weight spars…
University of Technology Sydney（悉尼科技大学）；Fudan University（复旦大学）；Hangdian University（杭州电子科技大学）
Deeper and wider Convolutional Neural Networks (CNNs) achieve superior performance but bring expensive computation cost. Accelerating such over-parameterized neural network has received increased attention. A typical pruning algorithm is a three-stage pipeline, i.e., training, pruning, and retraining. Prevailing approaches fix the pruned filters to zero during retraining, and thus significantly reduce the optimization space. Besides, they directly prune a large number of filters at first, which would cause unrecoverable information loss. To solve these problems, we propose an Asymptotic Soft F…
Harbin Institute of Technology（哈尔滨工业大学）；Peng Cheng Laboratory（鹏城实验室）；Tsinghua University（清华大学）
Guided filter is a fundamental tool in computer vision and computer graphics which aims to transfer structure information from guidance image to target image. Most existing methods construct filter kernels from the guidance itself without considering the mutual dependency between the guidance and the target. However, since there typically exist significantly different edges in the two images, simply transferring all structural information of the guidance to the target would result in various artifacts. To cope with this problem, we propose an effective framework named deep attentional guided i…
National Key Laboratory for Novel Software Technology（新型软件技术国家重点实验室）；Nanjing University（南京大学）；Nanjing（南京）；China（中国）
Channel pruning is an important family of methods to speed up deep model's inference. Previous filter pruning algorithms regard channel pruning and model fine-tuning as two independent steps. This paper argues that combining them into a single end-to-end trainable system will lead to better results. We propose an efficient channel selection layer, namely AutoPruner, to find less important filters automatically in a joint training manner. Our AutoPruner takes previous activation responses as an input and generates a true binary index code for pruning. Hence, all the filters corresponding to zer…
The University of Sydney（悉尼大学）；SenseTime（商汤科技）；Tsinghua University（清华大学）；Tianjin University（天津大学）；Institute for Artificial Intelligence, Tsinghua University (THUAI)（清华大学人工智能研究院）；Beijing National Research Center for Information Science and Technology (BNRist)（北京信息科学与技术国家研究中心）
To deploy a well-trained CNN model on low-end computation edge devices, it is usually supposed to compress or prune the model under certain computation budget (e.g., FLOPs). Current filter pruning methods mainly leverage feature maps to generate important scores for filters and prune those with smaller scores, which ignores the variance of input batches to the difference in sparse structure over filters. In this paper, we propose a data agnostic filter pruning method that uses an auxiliary network named Dagger module to induce pruning and takes pretrained weights as input to learn the importan…
National Institute of Technology Silchar（西尔查尔国家理工学院）
Malicious URL detection is an emerging research area due to continuous modernization of various systems, for instance, Edge Computing. In this article, we present a novel malicious URL detection technique, called deepBF (deep learning and Bloom Filter). deepBF is presented in two-fold. Firstly, we propose a learned Bloom Filter using 2-dimensional Bloom Filter. We experimentally decide the best non-cryptography string hash function. Then, we derive a modified non-cryptography string hash function from the selected hash function for deepBF by introducing biases in the hashing method and compare…
University of California, Merced（加州大学默塞德分校）；Virginia Tech（弗吉尼亚理工大学）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）
Joint image filters leverage the guidance image as a prior and transfer the structural details from the guidance image to the target image for suppressing noise or enhancing spatial resolution. Existing methods either rely on various explicit filter constructions or hand-designed objective functions, thereby making it difficult to understand, improve, and accelerate these filters in a coherent framework. In this paper, we propose a learning-based approach for constructing joint filters based on Convolutional Neural Networks. In contrast to existing methods that consider only the guidance image…
Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences（中国科学院深圳先进技术研究院）；Tsinghua University（清华大学）
Deep neural networks have enjoyed remarkable success for various vision tasks, however it remains challenging to apply CNNs to domains lacking a regular underlying structures such as 3D point clouds. Towards this we propose a novel convolutional architecture, termed SpiderCNN, to efficiently extract geometric features from point clouds. SpiderCNN is comprised of units called SpiderConv, which extend convolutional operations from regular grids to irregular point sets that can be embedded in R^n, by parametrizing a family of convolutional filters. We design the filter as a product of a simple st…
To address memory and computation resource limitations for hardware-oriented acceleration of deep convolutional neural networks (CNNs), we present a computation flow, stacked filters stationary flow (SFS), and a corresponding data encoding format, relative indexed compressed sparse filter format (CSF), to make the best of data sparsity, and simplify data handling at execution time. And we also propose a three dimensional Single Instruction Multiple Data (3D-SIMD) processor architecture to illustrate how to accelerate deep CNNs by taking advantage of SFS flow and CSF format. Comparing with the …
University of Technology Sydney（悉尼科技大学）；CETC（中国电子科技集团公司）；Huawei（华为）；Baidu Research（百度研究院）
Previous works utilized ''smaller-norm-less-important'' criterion to prune filters with smaller norm values in a convolutional neural network. In this paper, we analyze this norm-based criterion and point out that its effectiveness depends on two requirements that are not always met: (1) the norm deviation of the filters should be large; (2) the minimum norm of the filters should be small. To solve this problem, we propose a novel filter pruning method, namely Filter Pruning via Geometric Median (FPGM), to compress the model regardless of those two requirements. Unlike previous methods, FPGM c…
Rochester Institute of Technology（罗切斯特理工学院）；Chester F. Carlson Center for Imaging Science（切斯特·F·卡尔森成像科学中心）；School of Mathematical Sciences, Rochester Institute of Technology（罗切斯特理工学院数学科学学院）
Hyperspectral imaging holds enormous potential to improve the state-of-the-art in aerial vehicle tracking with low spatial and temporal resolutions. Recently, adaptive multi-modal hyperspectral sensors have attracted growing interest due to their ability to record extended data quickly from aerial platforms. In this study, we apply popular concepts from traditional object tracking, namely (1) Kernelized Correlation Filters (KCF) and (2) Deep Convolutional Neural Network (CNN) features to aerial tracking in hyperspectral domain. We propose the Deep Hyperspectral Kernelized Correlation Filter ba…
To increase the flexibility and scalability of deep neural networks for image reconstruction, a framework is proposed based on bandpass filtering. For many applications, sensing measurements are performed indirectly. For example, in magnetic resonance imaging, data are sampled in the frequency domain. The introduction of bandpass filtering enables leveraging known imaging physics while ensuring that the final reconstruction is consistent with actual measurements to maintain reconstruction accuracy. We demonstrate this flexible architecture for reconstructing subsampled datasets of MRI scans. T…
Veermata Jijabai Technological Institute（维尔马塔·吉贾拜技术学院）；Indian Institute of Technology Gandhinagar（印度理工学院甘地讷格尔分校）
Removing blur caused by camera shake in images has always been a challenging problem in computer vision literature due to its ill-posed nature. Motion blur caused due to the relative motion between the camera and the object in 3D space induces a spatially varying blurring effect over the entire image. In this paper, we propose a novel deep filter based on Generative Adversarial Network (GAN) architecture integrated with global skip connection and dense architecture in order to tackle this problem. Our model, while bypassing the process of blur kernel estimation, significantly reduces the test …
Academy for Advanced Interdisciplinary Studies, Peking University（北京大学前沿交叉学科研究院）；Alibaba Group（阿里巴巴集团）
Image color harmonization algorithm aims to automatically match the color distribution of foreground and background images captured in different conditions. Previous deep learning based models neglect two issues that are critical for practical applications, namely high resolution (HR) image processing and model comprehensibility. In this paper, we propose a novel Deep Comprehensible Color Filter (DCCF) learning framework for high-resolution image harmonization. Specifically, DCCF first downsamples the original input image to its low-resolution (LR) counter-part, then learns four human comprehe…
University of California, San Diego（加州大学圣地亚哥分校）；University of Central Florida（中佛罗里达大学）
There is an increasing number of pre-trained deep neural network models. However, it is still unclear how to effectively use these models for a new task. Transfer learning, which aims to transfer knowledge from source tasks to a target task, is an effective solution to this problem. Fine-tuning is a popular transfer learning technique for deep neural networks where a few rounds of training are applied to the parameters of a pre-trained model to adapt them to a new task. Despite its popularity, in this paper, we show that fine-tuning suffers from several drawbacks. We propose an adaptive fine-t…
Deep learning has greatly improved visual recognition in recent years. However, recent research has shown that there exist many adversarial examples that can negatively impact the performance of such an architecture. This paper focuses on detecting those adversarial examples by analyzing whether they come from the same distribution as the normal examples. Instead of directly training a deep neural network to detect adversarials, a much simpler approach was proposed based on statistics on outputs from convolutional layers. A cascade classifier was designed to efficiently detect adversarials. Fu…
Indian Institute of Technology Madras（印度理工学院马德拉斯分校）
Under Display Cameras present a promising opportunity for phone manufacturers to achieve bezel-free displays by positioning the camera behind semi-transparent OLED screens. Unfortunately, such imaging systems suffer from severe image degradation due to light attenuation and diffraction effects. In this work, we present Deep Atrous Guided Filter (DAGF), a two-stage, end-to-end approach for image restoration in UDC systems. A Low-Resolution Network first restores image quality at low-resolution, which is subsequently used by the Guided Filter Network as a filtering input to produce a high-resolu…
TELUQ University（加拿大远程大学（TELUQ大学））；University of Bahrain（巴林大学）
During the last years, deep learning trackers achieved stimulating results while bringing interesting ideas to solve the tracking problem. This progress is mainly due to the use of learned deep features obtained by training deep convolutional neural networks (CNNs) on large image databases. But since CNNs were originally developed for image classification, appearance modeling provided by their deep layers might be not enough discriminative for the tracking task. In fact,such features represent high-level information, that is more related to object category than to a specific instance of the ob…
Indian Institute of Information Technology Sri City（斯里城印度信息技术学院）；Indian Institute of Information Technology, Allahabad（阿拉哈巴德印度信息技术学院）；Shiv Nadar University（希夫·纳达尔大学）
Deep Convolutional Neural Networks (DCNNs) have shown promising performances in several visual recognition problems which motivated the researchers to propose popular architectures such as LeNet, AlexNet, VGGNet, ResNet, and many more. These architectures come at a cost of high computational complexity and parameter storage. To get rid of storage and computational complexity, deep model compression methods have been evolved. We propose a "History Based Filter Pruning (HBFP)" method that utilizes network training history for filter pruning. Specifically, we prune the redundant filters by observ…
Center for Research in Computer Vision（计算机视觉研究中心）；Department of Computer Science（计算机科学系）；University of Central Florida（中佛罗里达大学）
We propose an efficient and straightforward method for compressing deep convolutional neural networks (CNNs) that uses basis filters to represent the convolutional layers, and optimizes the performance of the compressed network directly in the basis space. Specifically, any spatial convolution layer of the CNN can be replaced by two successive convolution layers: the first is a set of three-dimensional orthonormal basis filters, followed by a layer of one-dimensional filters that represents the original spatial filters in the basis space. We jointly fine-tune both the basis and the filter repr…
This paper examines the possibility of, and the possible advantages to learning the filters of convolutional neural networks (CNNs) for image analysis in the wavelet domain. We are stimulated by both Mallat's scattering transform and the idea of filtering in the Fourier domain. It is important to explore new spaces in which to learn, as these may provide inherent advantages that are not available in the pixel space. However, the scattering transform is limited by its inability to learn in between scattering orders, and any Fourier domain filtering is limited by the large number of filter param…
We propose an efficient and unified framework, namely ThiNet, to simultaneously accelerate and compress CNN models in both training and inference stages. We focus on the filter level pruning, i.e., the whole filter would be discarded if it is less important. Our method does not change the original network structure, thus it can be perfectly supported by any off-the-shelf deep learning libraries. We formally establish filter pruning as an optimization problem, and reveal that we need to prune filters based on statistics information computed from its next layer, not the current layer, which diff…
Interpretability of Deep Neural Networks has become a major area of exploration. Although these networks have achieved state of the art accuracy in many tasks, it is extremely difficult to interpret and explain their decisions. In this work we analyze the final and penultimate layers of Deep Convolutional Networks and provide an efficient method for identifying subsets of features that contribute most towards the network's decision for a class. We demonstrate that the number of such features per class is much lower in comparison to the dimension of the final layer and therefore the decision su…
University of Ljubljana（卢布尔雅那大学）；University of Birmingham（伯明翰大学）
Convolutional neural networks excel in a number of computer vision tasks. One of their most crucial architectural elements is the effective receptive field size, that has to be manually set to accommodate a specific task. Standard solutions involve large kernels, down/up-sampling and dilated convolutions. These require testing a variety of dilation and down/up-sampling factors and result in non-compact representations and excessive number of parameters. We address this issue by proposing a new convolution filter composed of displaced aggregation units (DAU). DAUs learn spatial displacements an…
In an effort to understand the meaning of the intermediate representations captured by deep networks, recent papers have tried to associate specific semantic concepts to individual neural network filter responses, where interesting correlations are often found, largely by focusing on extremal filter responses. In this paper, we show that this approach can favor easy-to-interpret cases that are not necessarily representative of the average behavior of a representation. A more realistic but harder-to-study hypothesis is that semantic representations are distributed, and thus filters must be stud…
