Fuzzy Choquet Integration of Deep Convolutional Neural Networks for Remote Sensing

Anderson, Derek T.; Scott, Grant; Islam, Muhammad Aminul; Murray, Bryce; Marcum, Richard A.

doi:10.1007/978-3-319-89629-8_1

Cited by 28 publications

(24 citation statements)

References 66 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Fusion strategies, notably those associated with fuzzy measures (FMs), are conventionally applied at the decision level to aggregate outputs and improve performance. For example, DL classifier outputs were fused in [16]- [18] to improve classification performance in remote sensing applications by deriving FMs for each class and then fusing the measures with the classifiers' outputs through either the Sugeno or Choquet integral (ChI) [19]. Still, fusion strategies occurring at the input level can also benefit classification performance.…”

Section: Related Workmentioning

confidence: 99%

Introducing Fuzzy Layers for Deep Learning

Price

Anderson

2019

2019 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE)

Self Cite

View full text Add to dashboard Cite

Many state-of-the-art technologies developed in recent years have been influenced by machine learning to some extent. Most popular at the time of this writing are artificial intelligence methodologies that fall under the umbrella of deep learning. Deep learning has been shown across many applications to be extremely powerful and capable of handling problems that possess great complexity and difficulty. In this work, we introduce a new layer to deep learning: the fuzzy layer. Traditionally, the network architecture of neural networks is composed of an input layer, some combination of hidden layers, and an output layer. We propose the introduction of fuzzy layers into the deep learning architecture to exploit the powerful aggregation properties expressed through fuzzy methodologies, such as the Choquet and Sugueno fuzzy integrals. To date, fuzzy approaches taken to deep learning have been through the application of various fusion strategies at the decision level to aggregate outputs from state-of-the-art pre-trained models, e.g., AlexNet, VGG16, GoogLeNet, Inception-v3, ResNet-18, etc. While these strategies have been shown to improve accuracy performance for image classification tasks, none have explored the use of fuzzified intermediate, or hidden, layers. Herein, we present a new deep learning strategy that incorporates fuzzy strategies into the deep learning architecture focused on the application of semantic segmentation using per-pixel classification. Experiments are conducted on a benchmark data set as well as a data set collected via an unmanned aerial system at a U.S. Army test site for the task of automatic road segmentation, and preliminary results are promising.

show abstract

Section: Related Workmentioning

confidence: 99%

Introducing Fuzzy Layers for Deep Learning

Price

Anderson

2019

2019 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE)

Self Cite

View full text Add to dashboard Cite

show abstract

“…The fuzzy integral has been demonstrated numerous times in a variety of applications; e.g., explosive hazard detection [22,23], computer vision [24], pattern recognition [25,26,27], multi-criteria decision making [28,29], forensic anthropology [30,31], fuzzy logic [32], multiple kernel learning [33], multiple instance learning [34], ontologies [35], missing data [36], and deep learning for remote sensing [37,38], to name a few. The ChI is a nonlinear aggregation function parameterized by the FM.…”

Section: Measure and Choquet Integralmentioning

confidence: 99%

“…The complete training procedure of how these networks were trained can be found in [46]. Furthermore, the complete description of how these DCNNs are fused can be found in [37]. With respect to each ChI, we can determine how many important sources there are for each class.…”

Section: Case Studymentioning

confidence: 99%

“…This results in distorted answers, i.e., results computed using variables whose information are not supported by data. To remedy this, we proposed ChI [46,37,54]. 6 In summary, while indices are essential to our ability to tackle XAI, they are low-level, they require expertise, they generally are associated with information overload, and ultimately they are far from what an end user would likely consider a quality high-level human explanation.…”

Section: Information Indicesmentioning

confidence: 99%

“…The ChI is an effective mechanism to aggregate data in many applications including explosive hazard detection [22,23], pattern recognition [25,27], multi-criteria decision making [28,29], fuzzy logic [32], and deep learning for remote sensing [37]. The Herein, we focus on how to transfer learn variables from another domain, e.g., a previously known model.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Explainable data fusion

Murray¹

View full text Add to dashboard Cite

The recent resurgence of Artificial Intelligence (AI), specifically in the context of applications like healthcare, security and defense, IoT, and other areas that have a big impact on human life, has led to a demand for eXplainable AI (XAI). The production of explanations is argued to be a key aspect of achieving goals like trustworthiness and transparent versus opaque AI. XAI is also of fundamental academic interest with respect to helping us identifying weaknesses in the pursuit of making better AI. Herein, I focus on one piece of the AI puzzle, information fusion. In this work, I propose XAI fusion indices, linguistic summaries (aka textual explanations) of these indices, and local explanations for the fuzzy integral. However, a limitation of these indices is its tailored to highly educated fusion experts, and it is not clear what to do with these explanations. Herein, I extend the introduced indices to actionable explanations, which are demonstrated in the context of two case studies; multi-source fusion and deep learning for remote sensing. This work ultimately shows what XAI for fusion is and how to create actionable insights.

show abstract