Evaluating Capability of Deep Neural Networks for Image Classification via Information Plane

Cheng, Hao; Lian, Dongze; Gao, Shenghua; Geng, Yanlin

doi:10.1007/978-3-030-01252-6_11

Cited by 26 publications

(50 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Another indicator to interpret CNN's performance and capability is by loss curve to show whether the optimization process and relative learning progress improves for several epochs [48]. As the classification process involves generalization, so there is some level of information loss which leads to loss of completeness in the final result [49]. The loss function is generally defined by:…”

Section: Evaluation Metricsmentioning

confidence: 99%

Land Cover Classification from fused DSM and UAV Images Using Convolutional Neural Networks

et al. 2019

View full text Add to dashboard Cite

In recent years, remote sensing researchers have investigated the use of different modalities (or combinations of modalities) for classification tasks. Such modalities can be extracted via a diverse range of sensors and images. Currently, there are no (or only a few) studies that have been done to increase the land cover classification accuracy via unmanned aerial vehicle (UAV)-digital surface model (DSM) fused datasets. Therefore, this study looks at improving the accuracy of these datasets by exploiting convolutional neural networks (CNNs). In this work, we focus on the fusion of DSM and UAV images for land use/land cover mapping via classification into seven classes: bare land, buildings, dense vegetation/trees, grassland, paved roads, shadows, and water bodies. Specifically, we investigated the effectiveness of the two datasets with the aim of inspecting whether the fused DSM yields remarkable outcomes for land cover classification. The datasets were: (i) only orthomosaic image data (Red, Green and Blue channel data), and (ii) a fusion of the orthomosaic image and DSM data, where the final classification was performed using a CNN. CNN, as a classification method, is promising due to hierarchical learning structure, regulating and weight sharing with respect to training data, generalization, optimization and parameters reduction, automatic feature extraction and robust discrimination ability with high performance. The experimental results show that a CNN trained on the fused dataset obtains better results with Kappa index of~0.98, an average accuracy of 0.97 and final overall accuracy of 0.98. Comparing accuracies between the CNN with DSM result and the CNN without DSM result for the overall accuracy, average accuracy and Kappa index revealed an improvement of 1.2%, 1.8% and 1.5%, respectively. Accordingly, adding the heights of features such as buildings and trees improved the differentiation between vegetation specifically where plants were dense.

show abstract

Section: Evaluation Metricsmentioning

confidence: 99%

Land Cover Classification from fused DSM and UAV Images Using Convolutional Neural Networks

et al. 2019

View full text Add to dashboard Cite

show abstract

“…We performed experiments under different training conditions to train the neural networks. Table 1 , from [ 15 ], records percentages with different training conditions.…”

Section: Main Resultsmentioning

confidence: 99%

“…Thus, we also validated our hypothesis on the validation data where X represents the validation input. Table 3 , from [ 15 ], shows that low

also contributed to the validation accuracy. This observation will form the basis of our evaluation framework proposed in the next section.…”

Section: Main Resultsmentioning

confidence: 99%

“…Specifically, they estimated the mutual information

and

for each hidden layer , where X is the input data, Y is the label, and T is the hidden layer output, respectively. Figure 1 , from [ 15 ], draws

with training epochs in the two-dimensional plane (referred to as the information plane). From the figure, the green points (transition points) separate mutual information paths into two stages: The first phase, which is also called the “fitting phase”, only takes a little time in the whole training procedure.…”

Section: Introductionmentioning

confidence: 99%

“…Compared to our original conference version [ 15 ], this paper focuses more on application scenarios arising in deep learning and image classification problems like transfer learning and on the comparison of different optimization algorithms listed above.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Utilizing Information Bottleneck to Evaluate the Capability of Deep Neural Networks for Image Classification

Cheng

Lian

Gao

et al. 2019

Entropy

Self Cite

View full text Add to dashboard Cite

Inspired by the pioneering work of the information bottleneck (IB) principle for Deep Neural Networks’ (DNNs) analysis, we thoroughly study the relationship among the model accuracy, I ( X ; T ) and I ( T ; Y ) , where I ( X ; T ) and I ( T ; Y ) are the mutual information of DNN’s output T with input X and label Y. Then, we design an information plane-based framework to evaluate the capability of DNNs (including CNNs) for image classification. Instead of each hidden layer’s output, our framework focuses on the model output T. We successfully apply our framework to many application scenarios arising in deep learning and image classification problems, such as image classification with unbalanced data distribution, model selection, and transfer learning. The experimental results verify the effectiveness of the information plane-based framework: Our framework may facilitate a quick model selection and determine the number of samples needed for each class in the unbalanced classification problem. Furthermore, the framework explains the efficiency of transfer learning in the deep learning area.

show abstract

DisCo: Remedying Self-supervised Learning on Lightweight Models with Distilled Contrastive Learning

Gao

Zhuang

Lin

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Evaluating Capability of Deep Neural Networks for Image Classification via Information Plane

Cited by 26 publications

References 7 publications

Land Cover Classification from fused DSM and UAV Images Using Convolutional Neural Networks

Land Cover Classification from fused DSM and UAV Images Using Convolutional Neural Networks

Utilizing Information Bottleneck to Evaluate the Capability of Deep Neural Networks for Image Classification

DisCo: Remedying Self-supervised Learning on Lightweight Models with Distilled Contrastive Learning

Contact Info

Product

Resources

About