Deep learning has emerged as a powerful tool for feature-driven labeling of datasets. However, for it to be effective, it requires a large and finely labeled training dataset. Precisely labeling a large training dataset is expensive, time-consuming, and error prone. In this article, we present a visually driven deep-learning approach that starts with a coarsely labeled training dataset and iteratively refines the labeling through intuitive interactions that leverage the latent structures of the dataset. Our approach can be used to (a) alleviate the burden of intensive manual labeling that captures the fine nuances in a high-dimensional dataset by simple visual interactions, (b) replace a complicated (and therefore difficult to design) labeling algorithm by a simpler (but coarse) labeling algorithm supplemented by user interaction to refine the labeling, or (c) use low-dimensional features (such as the RGB colors) for coarse labeling and turn to higher-dimensional latent structures that are progressively revealed by deep learning, for fine labeling. We validate our approach through use cases on three high-dimensional datasets and a user study.

Designing volume visualizations showing various structures of interest is critical to the exploratory analysis of volumetric data. The last few years have witnessed dramatic advances in the use of convolutional neural networks for identification of objects in large image collections. Whereas such machine learning methods have shown superior performance in a number of applications, their direct use in volume visualization has not yet been explored. In this paper, we present a deep-learning-assisted volume visualization to depict complex structures, which are otherwise challenging for conventional approaches. A significant challenge in designing volume visualizations based on the high-dimensional deep features lies in efficiently handling the immense amount of information that deep-learning methods provide. In this paper, we present a new technique that uses spectral methods to facilitate user interactions with high-dimensional features. We also present a new deep-learning-assisted technique for hierarchically exploring a volumetric dataset. We have validated our approach on two electron microscopy volumes and one magnetic resonance imaging dataset.

Analyzing live-cell images is particularly challenging because cells simultaneously move and undergo systematic changes. Visually inspecting live-cell images therefore involves simultaneously tracking individual cells and detecting relevant spatio-temporal changes. The high cognitive burden of such a complex task makes this kind of analysis inefficient and error prone. In this paper, we describe a deep-learning-assisted visualization based on automatically derived high-level features to identify target cell changes in live-cell images. Applying a novel user-mediated color assignment scheme that maps abstract features into corresponding colors, we create color-based visual annotations that facilitate visual reasoning and analysis of complex time-varying live-cell image datasets.

Modern imaging technologies enable the study of microstructural features, which require capturing the finest details in high-resolution gigapixel images. Nevertheless, the resolution disparity between gigapixel images and megapixel displays presents a challenge to effective visual analysis because subtle texture differences are hardly perceivable at coarser resolutions. In this paper, we present a hierarchical segmentation technique based on the joint distribution of intensity and noise-resistant local binary patterns to differentiate subtle microstructural textures across various scales. The coarse-to-fine segmentation procedure subdivides each parent segment into texturally-distinct child segments at progressively higher resolutions. The hierarchical structure of segments allows creating intermediate segmentation results interactively. Based on the intermediate results, we highlight regions with texture differences using distinct colors, which provide salient visual hints to users despite the current viewing resolution. Our new technique has been validated on large microscopy images and shows promising results.

Much of the success of convolutional neural networks (CNNs) is due to the enormous collections of labeled data, powerful GPUs, and modern network architectures that facilitate the training and testing of larger and deeper models. The limited size of labeled volumetric microscopy images, however, hinders the proper training of CNNs for volume segmentation. Here, we design various 2D and 3D CNNs with factorized convolutions and online feature-level augmentations to address challenges due to the scarcity of training data. Based on the experimental results, we found that the 3D CNNs consistently outperformed the 2D counterparts. In addition, the 3D CNN that uses both factorized convolutions and online feature-level augmentations achieved the best segmentation performance.

The multiobjective flowshop problem with makespan and total flow time as objectives is addressed. A genetic local search algorithm is proposed with the ability to allocate the computational resources through the dynamic population size and local search intensity. The proposed method is compared with existing algorithms for flowshop scheduling with a public benchmark problem set. The experimental results show that the proposed method is capable of discovering solutions with better quality and diversity. The proposed method yields the best known nondominated solutions for the commonly studied permutation flowshop benchmarks, and the set of best known solutions is useful for the evaluation of performance of future studies.

The identical parallel batch machine scheduling problem is addressed in this paper. Incompatible job families and dynamic job arrivals are considered, and the objective is to minimize total weighted tardiness. A memetic algorithm is proposed to assign the batches to machines and to determine their processing sequences. The proposed approach is shown to outperform an existing approach in terms of solution quality and computational efficiency through comprehensive experiments.

In this paper, we solve two problems in a 300-mm wafer fabrication facility (fab). Firstly, for the lot scheduling problem, we propose a multi-objective genetic programming based rule generator (MOGPRG) to evolve useful dispatching rules, which can provide near-optimal lot schedules concerning multiple objectives. Secondly, the overhead hoist transports (OHT) routing problem is considered. As the modern automated material handling system (AMHS) is capable of doing tool-to-tool direct delivery, the congestion of OHTs may happen more often than the past. To deal with the traffic congestion in AMHS, a dynamic routing method is proposed to find the near-shortest and less-congested path for the OHT to travel along. It can reduce the traffic congestion and achieve fast lot delivery by adapting to the dynamic traffic environment. The proposed MOGPRG is integrated with the dynamic routing method to improve two fab performance metrics: mean cycle time and tardy rate. Experimental results show the effectiveness of the proposed MOGPRG and dynamic routing method.