Methods and Metrics for Explaining Artificial Intelligence Models: A Review

Banerjee, Puja; Barnwal, Rajesh P.

doi:10.1007/978-3-031-12807-3_4

Cited by 4 publications

(4 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Drusen has slightly lower sensitivity, while CNV has the highest specificity (>0.97), suggesting strong performance in distinguishing its class. Because it is hard to understand how the CNN model predicted the output, XAI techniques are used to explain it [32]. The testing images are on the left, and each explanation has a transparent grey background (see .…”

Section: Statistical Resultsmentioning

confidence: 99%

“…The proposed lightweight model reported in Table 1 provides a concise overview of the CNN architecture and Figure 4 shows the graphical result. The model holds only five convolution layers to perform convolution operations on the input image, with increasing filter depths (16,32,64,128,256) to capture hierarchical features. Each convolution layer is followed by max-pooling layers to down-sample the feature maps, aiding in information compression.…”

Section: Cnn Modelmentioning

confidence: 99%

See 1 more Smart Citation

Evaluating Retinal Disease Diagnosis with an Interpretable Lightweight CNN Model Resistant to Adversarial Attacks

Bhandari,

Shahi,

Neupane

2023

J. Imaging

View full text Add to dashboard Cite

Optical Coherence Tomography (OCT) is an imperative symptomatic tool empowering the diagnosis of retinal diseases and anomalies. The manual decision towards those anomalies by specialists is the norm, but its labor-intensive nature calls for more proficient strategies. Consequently, the study recommends employing a Convolutional Neural Network (CNN) for the classification of OCT images derived from the OCT dataset into distinct categories, including Choroidal NeoVascularization (CNV), Diabetic Macular Edema (DME), Drusen, and Normal. The average k-fold (k = 10) training accuracy, test accuracy, validation accuracy, training loss, test loss, and validation loss values of the proposed model are 96.33%, 94.29%, 94.12%, 0.1073, 0.2002, and 0.1927, respectively. Fast Gradient Sign Method (FGSM) is employed to introduce non-random noise aligned with the cost function’s data gradient, with varying epsilon values scaling the noise, and the model correctly handles all noise levels below 0.1 epsilon. Explainable AI algorithms: Local Interpretable Model-Agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP) are utilized to provide human interpretable explanations approximating the behaviour of the model within the region of a particular retinal image. Additionally, two supplementary datasets, namely, COVID-19 and Kidney Stone, are assimilated to enhance the model’s robustness and versatility, resulting in a level of precision comparable to state-of-the-art methodologies. Incorporating a lightweight CNN model with 983,716 parameters, 2.37×108 floating point operations per second (FLOPs) and leveraging explainable AI strategies, this study contributes to efficient OCT-based diagnosis, underscores its potential in advancing medical diagnostics, and offers assistance in the Internet-of-Medical-Things.

show abstract

Section: Statistical Resultsmentioning

confidence: 99%

Section: Cnn Modelmentioning

confidence: 99%

Evaluating Retinal Disease Diagnosis with an Interpretable Lightweight CNN Model Resistant to Adversarial Attacks

Bhandari,

Shahi,

Neupane

2023

J. Imaging

View full text Add to dashboard Cite

show abstract

“…Since it was difficult to directly interpret the mathematical behavior of the CNN model, XAI techniques were applied to the model [60]. Four results were illustrated by the SHAP results for each category (cyst, normal, stone, tumor).Testing images are shown on the left with a transparent gray background behind each explanation.…”

Section: Descriptive Analysis From Xai 521 Shapmentioning

confidence: 99%

Exploring the Capabilities of a Lightweight CNN Model in Accurately Identifying Renal Abnormalities: Cysts, Stones, and Tumors, Using LIME and SHAP

et al. 2023

View full text Add to dashboard Cite

Kidney abnormality is one of the major concerns in modern society, and it affects millions of people around the world. To diagnose different abnormalities in human kidneys, a narrow-beam x-ray imaging procedure, computed tomography, is used, which creates cross-sectional slices of the kidneys. Several deep-learning models have been successfully applied to computer tomography images for classification and segmentation purposes. However, it has been difficult for clinicians to interpret the model’s specific decisions and, thus, creating a “black box” system. Additionally, it has been difficult to integrate complex deep-learning models for internet-of-medical-things devices due to demanding training parameters and memory-resource cost. To overcome these issues, this study proposed (1) a lightweight customized convolutional neural network to detect kidney cysts, stones, and tumors and (2) understandable AI Shapely values based on the Shapley additive explanation and predictive results based on the local interpretable model-agnostic explanations to illustrate the deep-learning model. The proposed CNN model performed better than other state-of-the-art methods and obtained an accuracy of 99.52 ± 0.84% for K = 10-fold of stratified sampling. With improved results and better interpretive power, the proposed work provides clinicians with conclusive and understandable results.

show abstract

“…Furthermore, the integration of models and methods from Explainable Artificial Intelligence (Banerjee and Barnwal, 2023), especially in the processes that involve Machine Learning algorithms (e.g., argument mining or argument generation) will contribute to the transparency, interpretability and understandability of the outputs of the Web of Debates tools and applications and to the establishment of trust with their users. Computational argumentation has already proved to be a very http://en.wikipedia.org/wiki/Fact_checking https://en.wikipedia.org/wiki/List_of_fact-checking_websites useful tool for developing explainable systems (Vassiliades et al, 2021), while the recent launch of the International Workshop on Argumentation for Explainable AI shows that this is an active area of interest for researchers in computational argumentation.…”

Section: Ethical Issuesmentioning

confidence: 99%

Sketching the vision of the Web of Debates

Bikakis

Flouris

Patkos

et al. 2023

Front. Artif. Intell.

View full text Add to dashboard Cite

The exchange of comments, opinions, and arguments in blogs, forums, social media, wikis, and review websites has transformed the Web into a modern agora, a virtual place where all types of debates take place. This wealth of information remains mostly unexploited: due to its textual form, such information is difficult to automatically process and analyse in order to validate, evaluate, compare, combine with other types of information and make it actionable. Recent research in Machine Learning, Natural Language Processing, and Computational Argumentation has provided some solutions, which still cannot fully capture important aspects of online debates, such as various forms of unsound reasoning, arguments that do not follow a standard structure, information that is not explicitly expressed, and non-logical argumentation methods. Tackling these challenges would give immense added-value, as it would allow searching for, navigating through and analyzing online opinions and arguments, obtaining a better picture of the various debates for a well-intentioned user. Ultimately, it may lead to increased participation of Web users in democratic, dialogical interchange of arguments, more informed decisions by professionals and decision-makers, as well as to an easier identification of biased, misleading, or deceptive arguments. This paper presents the vision of the Web of Debates, a more human-centered version of the Web, which aims to unlock the potential of the abundance of argumentative information that currently exists online, offering its users a new generation of argument-based web services and tools that are tailored to their real needs.

show abstract

Methods and Metrics for Explaining Artificial Intelligence Models: A Review

Cited by 4 publications

References 31 publications

Evaluating Retinal Disease Diagnosis with an Interpretable Lightweight CNN Model Resistant to Adversarial Attacks

Evaluating Retinal Disease Diagnosis with an Interpretable Lightweight CNN Model Resistant to Adversarial Attacks

Exploring the Capabilities of a Lightweight CNN Model in Accurately Identifying Renal Abnormalities: Cysts, Stones, and Tumors, Using LIME and SHAP

Sketching the vision of the Web of Debates

Contact Info

Product

Resources

About