A pseudo-softmax function for hardware-based high speed image classification

AbstractIn this work a novel architecture, named pseudo-softmax, to compute an approximated form of the softmax function is presented. This architecture can be fruitfully used in the last layer of Neural Networks and Convolutional Neural Networks for classification tasks, and in Reinforcement Learning hardware accelerators to compute the Boltzmann action-selection policy. The proposed pseudo-softmax design, intended for efficient hardware implementation, exploits the typical integer quantization of hardware-based Neural Networks obtaining an accurate approximation of the result. In the paper, a detailed description of the architecture is given and an extensive analysis of the approximation error is performed by using both custom stimuli and real-world Convolutional Neural Networks inputs. The implementation results, based on CMOS standard-cell technology, compared to state-of-the-art architectures show reduced approximation errors.

Download Full-text

Automatic Tool for Fast Generation of Custom Convolutional Neural Networks Accelerators for FPGA

Electronics ◽

10.3390/electronics8060641 ◽

2019 ◽

Vol 8 (6) ◽

pp. 641 ◽

Cited By ~ 7

Author(s):

Miguel Rivera-Acosta ◽

Susana Ortega-Cisneros ◽

Jorge Rivera

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

State Of The Art ◽

Third Party ◽

Hardware Accelerators ◽

Field Programmable ◽

Custom Hardware ◽

Architecture Description ◽

Automatic Tool

This paper presents a platform that automatically generates custom hardware accelerators for convolutional neural networks (CNNs) implemented in field-programmable gate array (FPGA) devices. It includes a user interface for configuring and managing these accelerators. The herein-presented platform can perform all the processes necessary to design and test CNN accelerators from the CNN architecture description at both layer and internal parameter levels, training the desired architecture with any dataset and generating the configuration files required by the platform. With these files, it can synthesize the register-transfer level (RTL) and program the customized CNN accelerator into the FPGA device for testing, making it possible to generate custom CNN accelerators quickly and easily. All processes save the CNN architecture description are fully automatized and carried out by the platform, which manages third-party software to train the CNN and synthesize and program the generated RTL. The platform has been tested with the implementation of some of the CNN architectures found in the state-of-the-art for freely available datasets such as MNIST, CIFAR-10, and STL-10.

Download Full-text

HierarchyNet: Hierarchical CNN-Based Urban Building Classification

Remote Sensing ◽

10.3390/rs12223794 ◽

2020 ◽

Vol 12 (22) ◽

pp. 3794

Author(s):

Salma Taoufiq ◽

Balázs Nagy ◽

Csaba Benedek

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

State Of The Art ◽

Feedback Signal ◽

Hierarchical Network ◽

Urban Buildings ◽

Virtual City ◽

Classification Tasks ◽

Coarse To Fine

Automatic building categorization and analysis are particularly relevant for smart city applications and cultural heritage programs. Taking a picture of the facade of a building and instantly obtaining information about it can enable the automation of processes in urban planning, virtual city tours, and digital archiving of cultural artifacts. In this paper, we go beyond traditional convolutional neural networks (CNNs) for image classification and propose the HierarchyNet: a new hierarchical network for the classification of urban buildings from all across the globe into different main and subcategories from images of their facades. We introduce a coarse-to-fine hierarchy on the dataset and the model learns to simultaneously extract features and classify across both levels of hierarchy. We propose a new multiplicative layer, which is able to improve the accuracy of the finer prediction by considering the feedback signal of the coarse layers. We have quantitatively evaluated the proposed approach both on our proposed building datasets, as well as on various benchmark databases to demonstrate that the model is able to efficiently learn hierarchical information. The HierarchyNet model is able to outperform the state-of-the-art convolutional neural networks in urban building classification as well as in other multi-label classification tasks while using significantly fewer parameters.

Download Full-text

A Neural N-Gram Network for Text Classification

Journal of Advanced Computational Intelligence and Intelligent Informatics ◽

10.20965/jaciii.2018.p0380 ◽

2018 ◽

Vol 22 (3) ◽

pp. 380-386 ◽

Cited By ~ 2

Author(s):

Zhenguo Yan ◽

◽

Yue Wu

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

Text Classification ◽

Word Order ◽

Input Data ◽

State Of The Art ◽

Poor Performance ◽

Proposed Model ◽

Classification Tasks ◽

N Gram

Convolutional Neural Networks (CNNs) effectively extract local features from input data. However, CNN based on word embedding and convolution layers displays poor performance in text classification tasks when compared with traditional baseline methods. We address this problem and propose a model named NNGN that simplifies the convolution layer in the CNN by replacing it with a pooling layer that extracts n-gram embedding in a simpler way and obtains document representations via linear computation. We implement two settings in our model to extract n-gram features. In the first setting, which we refer to as seq-NNGN, we consider word order within each n-gram. In the second setting, BoW-NNGN, we do not consider word order. We compare the performance of these settings in different classification tasks with those of other models. The experimental results show that our proposed model achieves better performance than state-of-the-art models.

Download Full-text

Using spatial-temporal ensembles of convolutional neural networks for lumen segmentation in ureteroscopy

International Journal of Computer Assisted Radiology and Surgery ◽

10.1007/s11548-021-02376-3 ◽

2021 ◽

Author(s):

Jorge F. Lazo ◽

Aldo Marzullo ◽

Sara Moccia ◽

Michele Catellani ◽

Benoit Rosa ◽

...

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

State Of The Art ◽

Automatic Segmentation ◽

Temporal Information ◽

Invasive Technique ◽

Dice Similarity Coefficient ◽

Specular Reflections ◽

Lumen Segmentation ◽

Previous State

Abstract Purpose Ureteroscopy is an efficient endoscopic minimally invasive technique for the diagnosis and treatment of upper tract urothelial carcinoma. During ureteroscopy, the automatic segmentation of the hollow lumen is of primary importance, since it indicates the path that the endoscope should follow. In order to obtain an accurate segmentation of the hollow lumen, this paper presents an automatic method based on convolutional neural networks (CNNs). Methods The proposed method is based on an ensemble of 4 parallel CNNs to simultaneously process single and multi-frame information. Of these, two architectures are taken as core-models, namely U-Net based in residual blocks ($$m_1$$ m 1 ) and Mask-RCNN ($$m_2$$ m 2 ), which are fed with single still-frames I(t). The other two models ($$M_1$$ M 1 , $$M_2$$ M 2 ) are modifications of the former ones consisting on the addition of a stage which makes use of 3D convolutions to process temporal information. $$M_1$$ M 1 , $$M_2$$ M 2 are fed with triplets of frames ($$I(t-1)$$ I ( t - 1 ) , I(t), $$I(t+1)$$ I ( t + 1 ) ) to produce the segmentation for I(t). Results The proposed method was evaluated using a custom dataset of 11 videos (2673 frames) which were collected and manually annotated from 6 patients. We obtain a Dice similarity coefficient of 0.80, outperforming previous state-of-the-art methods. Conclusion The obtained results show that spatial-temporal information can be effectively exploited by the ensemble model to improve hollow lumen segmentation in ureteroscopic images. The method is effective also in the presence of poor visibility, occasional bleeding, or specular reflections.

Download Full-text

Hybrid Mamdani Fuzzy Rules and Convolutional Neural Networks for Analysis and Identification of Animal Images

Computation ◽

10.3390/computation9030035 ◽

2021 ◽

Vol 9 (3) ◽

pp. 35

Author(s):

Hind R. Mohammed ◽

Zahir M. Hussain

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

High Speed ◽

Moving Objects ◽

Real Life ◽

Fuzzy Rules ◽

High Rate ◽

Moving Images ◽

Animal Images

Accurate, fast, and automatic detection and classification of animal images is challenging, but it is much needed for many real-life applications. This paper presents a hybrid model of Mamdani Type-2 fuzzy rules and convolutional neural networks (CNNs) applied to identify and distinguish various animals using different datasets consisting of about 27,307 images. The proposed system utilizes fuzzy rules to detect the image and then apply the CNN model for the object’s predicate category. The CNN model was trained and tested based on more than 21,846 pictures of animals. The experiments’ results of the proposed method offered high speed and efficiency, which could be a prominent aspect in designing image-processing systems based on Type 2 fuzzy rules characterization for identifying fixed and moving images. The proposed fuzzy method obtained an accuracy rate for identifying and recognizing moving objects of 98% and a mean square error of 0.1183464 less than other studies. It also achieved a very high rate of correctly predicting malicious objects equal to recall = 0.98121 and a precision rate of 1. The test’s accuracy was evaluated using the F1 Score, which obtained a high percentage of 0.99052.

Download Full-text

Image Classification for the Automatic Feature Extraction in Human Worn Fashion Data

Mathematics ◽

10.3390/math9060624 ◽

2021 ◽

Vol 9 (6) ◽

pp. 624

Author(s):

Stefan Rohrmanstorfer ◽

Mikhail Komarov ◽

Felix Mödritscher

Keyword(s):

Neural Networks ◽

Feature Extraction ◽

Image Classification ◽

Convolutional Neural Networks ◽

Data Augmentation ◽

State Of The Art ◽

Image Data ◽

Classification Model ◽

Upper Body ◽

Automatic Feature Extraction

With the always increasing amount of image data, it has become a necessity to automatically look for and process information in these images. As fashion is captured in images, the fashion sector provides the perfect foundation to be supported by the integration of a service or application that is built on an image classification model. In this article, the state of the art for image classification is analyzed and discussed. Based on the elaborated knowledge, four different approaches will be implemented to successfully extract features out of fashion data. For this purpose, a human-worn fashion dataset with 2567 images was created, but it was significantly enlarged by the performed image operations. The results show that convolutional neural networks are the undisputed standard for classifying images, and that TensorFlow is the best library to build them. Moreover, through the introduction of dropout layers, data augmentation and transfer learning, model overfitting was successfully prevented, and it was possible to incrementally improve the validation accuracy of the created dataset from an initial 69% to a final validation accuracy of 84%. More distinct apparel like trousers, shoes and hats were better classified than other upper body clothes.

Download Full-text

Canopy Height Estimation at Landsat Resolution Using Convolutional Neural Networks

Machine Learning and Knowledge Extraction ◽

10.3390/make2010003 ◽

2020 ◽

Vol 2 (1) ◽

pp. 23-36

Author(s):

Syed Aamir Ali Shah ◽

Muhammad Asif Manzoor ◽

Abdul Bais

Keyword(s):

Neural Networks ◽

Random Forest ◽

Convolutional Neural Networks ◽

Forest Structure ◽

State Of The Art ◽

Spatial Association ◽

Canopy Height ◽

Landsat Images ◽

Structure Estimation ◽

Vegetation Height

Forest structure estimation is very important in geological, ecological and environmental studies. It provides the basis for the carbon stock estimation and effective means of sequestration of carbon sources and sinks. Multiple parameters are used to estimate the forest structure like above ground biomass, leaf area index and diameter at breast height. Among all these parameters, vegetation height has unique standing. In addition to forest structure estimation it provides the insight into long term historical changes and the estimates of stand age of the forests as well. There are multiple techniques available to estimate the canopy height. Light detection and ranging (LiDAR) based methods, being the accurate and useful ones, are very expensive to obtain and have no global coverage. There is a need to establish a mechanism to estimate the canopy height using freely available satellite imagery like Landsat images. Multiple studies are available which contribute in this area. The majority use Landsat images with random forest models. Although random forest based models are widely used in remote sensing applications, they lack the ability to utilize the spatial association of neighboring pixels in modeling process. In this research work, we define Convolutional Neural Network based model and analyze that model for three test configurations. We replicate the random forest based setup of Grant et al., which is a similar state-of-the-art study, and compare our results and show that the convolutional neural networks (CNN) based models not only capture the spatial association of neighboring pixels but also outperform the state-of-the-art.

Download Full-text

Segment convolutional neural networks (Seg-CNNs) for classifying relations in clinical notes

Journal of the American Medical Informatics Association ◽

10.1093/jamia/ocx090 ◽

2017 ◽

Vol 25 (1) ◽

pp. 93-98 ◽

Cited By ~ 31

Author(s):

Yuan Luo ◽

Yu Cheng ◽

Özlem Uzuner ◽

Peter Szolovits ◽

Justin Starren

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

State Of The Art ◽

Graphics Processing Unit ◽

Medical Problem ◽

Feature Engineering ◽

Processing Unit ◽

Clinical Notes ◽

Overall Evaluation ◽

Relation Classification

Abstract We propose Segment Convolutional Neural Networks (Seg-CNNs) for classifying relations from clinical notes. Seg-CNNs use only word-embedding features without manual feature engineering. Unlike typical CNN models, relations between 2 concepts are identified by simultaneously learning separate representations for text segments in a sentence: preceding, concept1, middle, concept2, and succeeding. We evaluate Seg-CNN on the i2b2/VA relation classification challenge dataset. We show that Seg-CNN achieves a state-of-the-art micro-average F-measure of 0.742 for overall evaluation, 0.686 for classifying medical problem–treatment relations, 0.820 for medical problem–test relations, and 0.702 for medical problem–medical problem relations. We demonstrate the benefits of learning segment-level representations. We show that medical domain word embeddings help improve relation classification. Seg-CNNs can be trained quickly for the i2b2/VA dataset on a graphics processing unit (GPU) platform. These results support the use of CNNs computed over segments of text for classifying medical relations, as they show state-of-the-art performance while requiring no manual feature engineering.

Download Full-text