Fuzzy Pooling

Abstract

Convolutional Neural Networks (CNNs) are artificial learning systems typically based on two operations: convolution, which implements feature extraction through filtering, and pooling, which implements dimensionality reduction. The impact of pooling in the classification performance of the CNNs has been highlighted in several previous works, and a variety of alternative pooling operators have been proposed. However, only a few of them tackle with the uncertainty that is naturally propagated from the input layer to the feature maps of the hidden layers through convolutions. In this paper we present a novel pooling operation based on (type-1) fuzzy sets to cope with the local imprecision of the feature maps, and we investigate its performance in the context of image classification. Fuzzy pooling is performed by fuzzification, aggregation and defuzzification of feature map neighborhoods. It is used for the construction of a fuzzy pooling layer that can be applied as a drop-in replacement of the current, crisp, pooling layers of CNN architectures. Several experiments using publicly available datasets show that the proposed approach can enhance the classification performance of a CNN. A comparative evaluation shows that it outperforms state-of-the-art pooling approaches.

Real-time echocardiography image analysis and quantification of cardiac indices

“…The results in Table 1 demonstrate that the use of fuzzy pooling out- performs other pooling operations in both tasks (view classification and quality assessment). We refer the interested reader to Diamantis and Iakovidis (2020) for detailed theoretical description of how the proposed fuzzy operation tackles the uncertainty (caused by speckle noise) naturally propagated from the input to the feature maps.…”

“…For the shared echo-specific encoder, we used the encoder part of the MobileNetV2-s autoencoder. As mentioned above, MobileNetV2-s encoder has only 5 inverted residual bottleneck blocks and a final fuzzy pooling layer ( Diamantis and Iakovidis, 2020 ) instead of average pooling layer. We then attached two heads, namely view classification head and quality assessment head, to MobileNetV2-s encoder as shown in Fig.…”

“…1 , a ). In the encoder part of the autoencoder, we used a smaller version of MobileNetV2 ( Sandler et al, 2018 ), which we call MobileNetV2-s. As compared to MobileNetV2 ( Sandler et al, 2018 ), MobileNetV2-s has only 5 inverted residual bottleneck blocks and a final fuzzy pooling layer ( Diamantis and Iakovidis, 2020 ) instead of average pooling layer. It has 55,620 parameters and is less than ≈ 2 megabytes (MB).…”

Particle Swarm Optimized Fuzzy CNN With Quantitative Feature Fusion for Ultrasound Image Quality Identification

“…Different studies have found that the ReLU activation function with the highest value of six ( 6) helps the network learn the sparse features [46][47]. Thus, we have selected the value of p and q based on the highest value (rmax).…”

ASML: Algorithm-Agnostic Architecture for Scalable Machine Learning

“…In most cases, it is also possible to deploy pre-trained models such as [3] and [4], as long as they are implemented in a supported framework. The flexibility of such services is limited, as while it is relatively easy to get started, it is difficult to efficiently incorporate ML models based on novel components, such as the fuzzy pooling layer proposed in [5], or complex ML-based data-processing pipelines, such as pipelines that include image preprocessing, integration of multiple heterogeneous ML algorithms with bidirectional data communication. Such pipelines are frequently met in state-ofthe-art pattern analysis applications spanning a variety of domains e.g., web content perception [6], obstacle detection and navigation for robotics [7] and assistive technologies [8], real-time analysis of medical image sequences during brain surgery [9] and gastrointestinal (GI) endoscopy [10].…”