In the context of deep learning research, where model introductions continually occur, the need for effective and efficient evaluation remains paramount. Existing methods often emphasize accuracy metrics, overlooking stability. To address this, the paper introduces the Accuracy-Stability Index (ASI), a quantitative measure incorporating both accuracy and stability for assessing deep learning models. Experimental results demonstrate the application of ASI, and a 3D surface model is presented for visualizing ASI, mean accuracy, and coefficient of variation. This paper addresses the important issue of quantitative benchmarking metrics for deep learning models, providing a new approach for accurately evaluating accuracy and stability of deep learning models. The paper concludes with discussions on potential weaknesses and outlines future research directions.

Image quality is important, and can affect overall performance in image processing and computer vision as well as for numerous other reasons. Image quality assessment (IQA) is consequently a vital task in different applications from aerial photography interpretation to object detection to medical image analysis. In previous research, the BRISQUE algorithm and the PSNR algorithm were evaluated with high resolution (≥ 512×384 pixels), but relatively small image sets (≤4,744 images). However, scientists have not evaluated IQA algorithms on low resolution (≤32×32 pixels), multi-perturbation, big image sets (for example, ≥60,000 different images not counting their perturbations). This study explores these two IQA algorithms through experimental investigation. We first chose two deep learning image sets, CIFAR-10 and MNIST. Then, we added 68 perturbations that add noise to the images in specific sequences and noise intensities. In addition, we tracked the performance outputs of the two IQA algorithms with singly and multiply noised images. After quantitatively analyzing experimental results, we report the limitations of the two IQAs with these noised CIFAR-10 and MNIST image sets. We also explain three potential root causes for performance degradation. These findings point out weaknesses of the two IQA algorithms. The research results provide guidance to scientists and engineers developing accurate, robust IQA algorithms. All source codes, related image sets, and figures are shared on the website (https://github.com/caperock/imagequality) to support future scientific and industrial projects.

Deep learning (DL) classifiers are often unstable in that they may change significantly when retested on perturbed images or low quality images. This paper adds to the fundamental body of work on the robustness of DL classifiers. We introduce a new two-dimensional benchmarking matrix to evaluate robustness of DL classifiers, and we also innovate a four-quadrant statistical visualization tool, including minimum accuracy, maximum accuracy, mean accuracy, and coefficient of variation, for benchmarking robustness of DL classifiers. To measure robust DL classifiers, we create comprehensive 69 benchmarking image sets, including a clean set, sets with single factor perturbations, and sets with two-factor perturbation conditions. After collecting experimental results, we first report that using two-factor perturbed images improves both robustness and accuracy of DL classifiers. The two-factor perturbation includes (1) two digital perturbations (salt & pepper noise and Gaussian noise) applied in both sequences, and (2) one digital perturbation (salt & pepper noise) and a geometric perturbation (rotation) applied in both sequences. All source codes, related image sets, and results are shared on the GitHub website at https://github.com/caperock/robustai to support future academic research and industry projects.

In this paper we discuss models for systems that support reading.Our account identifies important models and presents a framework for organizing them.To evaluate this account, we show its ability to suggest a wide range of text presentation systems, many of them novel.This evaluation not only provides interesting ideas for future systems, it also shows the usefulness of the account, and further, exemplifies a general approach to evaluating meta-level discussions such as this, namely, evaluating by assessing ability to generate interesting implications.

The ability to discriminate and distinguish among individual documents in the ever-increasing volume of information available through digital networks is becoming more and more difficult. With websites being added to the 100 million installed base by tens of thousands per month, information overload is inevitable (H. Berghel, 1997). There are two basic paradigms for dealing with this information overload: filtering (Information filtering, 1992) information before it reaches the end-user, and customizing the information after it arrives (Berleant & Berghel, 1994a, 1994b). Filtering remains primarily a server side activity since filtering at the client-side would necessitate unnecessary downloads. Information customization is a client-side activity designed to pick up where information filtering leaves off. In this article, we describe our vision of information customization and, along the way, chronicle the development of our proof-of-concept prototype, Cyberbrowser, for customizing information on the Web.