Deep Learning-based computational pathology algorithms have demonstrated profound ability to excel in a wide array of tasks that range from characterization of well known morphological phenotypes to predicting non human-identifiable features from histology such as molecular alterations. However, the development of robust, adaptable and accurate deep learning-based models often rely on the collection and time-costly curation large high-quality annotated training data that should ideally come from diverse sources and patient populations to cater for the heterogeneity that exists in such datasets. Multi-centric and collaborative integration of medical data across multiple institutions can naturally help overcome this challenge and boost the model performance but is limited by privacy concerns among other difficulties that may arise in the complex data sharing process as models scale towards using hundreds of thousands of gigapixel whole slide images. In this paper, we introduce privacy-preserving federated learning for gigapixel whole slide images in computational pathology using weakly-supervised attention multiple instance learning and differential privacy. We evaluated our approach on two different diagnostic problems using thousands of histology whole slide images with only slide-level labels. Additionally, we present a weakly-supervised learning framework for survival prediction and patient stratification from whole slide images and demonstrate its effectiveness in a federated setting. Our results show that using federated learning, we can effectively develop accurate weakly-supervised deep learning models from distributed data silos without direct data sharing and its associated complexities, while also preserving differential privacy using randomized noise generation. We also make available an easy-to-use federated learning for computational pathology software package: http://github.com/mahmoodlab/HistoFL.

This article demonstrates the importance of integrating the Internet-of-Things (IoT) devices and technologies for efficient energy management in a building environment. An Elman recurrent neural network (RNN) model and an exponential model are developed for electric energy consumption prediction in an IoT-driven building environment. The models predict electric energy consumption by electric loads in the near future by: 1) recognizing the existence of a relationship between the net electric energy consumption of the building's electric loads and the ambient temperature along with the occupancy state of the building and 2) employing the detected relationship to predict electric energy consumption using the forecasted temperature and the scheduled occupancy state. The building environment under consideration is the Real-Time Power and Intelligent Systems (RTPIS) laboratory integrated with intelligent monitoring and control capabilities using IoT devices and technologies. The electric loads under consideration include heating, ventilation, and air conditioning (HVAC) units and light panels. The developed Elman RNN and exponential models are also compared. These prediction capabilities are beneficial in overcoming variabilities in electric energy consumption by supplying electric energy as needed to meet the demands of the electric loads, thereby minimizing wasted electric energy, reducing carbon emissions, and generating cost savings.

Due to the importance of nuclear structure in cancer diagnosis, several predictive models have been described for diagnosing a wide variety of cancers based on nuclear morphology. In many computer-aided diagnosis (CAD) systems, cancer detection tasks can be generally formulated as set classification problems, which can not be directly solved by classifying single instances. In this paper, we propose a novel set classification approach SetSVM to build a predictive model by considering any nuclei set as a whole without specific assumptions. SetSVM features highly discriminative power in cancer detection challenges in the sense that it not only optimizes the classifier decision boundary but also transfers discriminative information to set representation learning. During model training, these two processes are unified in the support vector machine (SVM) maximum separation margin problem. Experiment results show that SetSVM provides significant improvements compared with five commonly used approaches in cancer detection tasks utilizing 260 patients in total across three different cancer types, namely, thyroid cancer, liver cancer, and melanoma. In addition, we show that SetSVM enables visual interpretation of discriminative nuclear characteristics representing the nuclei set. These features make SetSVM a potentially practical tool in building accurate and interpretable CAD systems for cancer detection.

A transformation is underway in electric power and energy systems (EPESs) to provide clean distributed energy for sustainable global economic growth. Internet of Things (IoT) is at the forefront of this transformation imparting capabilities, such as real-time monitoring, situational awareness and intelligence, control, and cyber security to transform the existing EPES into intelligent cyber-enabled EPES, which is more efficient, secure, reliable, resilient, and sustainable. Additionally, digitizing the electric power ecosystem using IoT improves asset visibility, optimal management of distributed generation, eliminates energy wastage, and create savings. IoT has a significant impact on EPESs and offers several opportunities for growth and development. There are several challenges with the deployment of IoT for EPESs. Viable solutions need to be developed to overcome these challenges to ensure continued growth of IoT for EPESs. The advancements in computational intelligence capabilities can evolve an intelligent IoT system by emulating biological nervous systems with cognitive computation, streaming and distributed analytics including at the edge and device levels. This review paper provides an assessment of the role, impact and challenges of IoT in transforming EPESs.

With the ongoing challenge of successfully managing information technology (IT) projects, organizations are recognizing the need for greater project management discipline. For many organizations, this has meant ratcheting up project management skills, processes, and governance structures by implementing a project management office (PMO). While anecdotal evidence suggests that implementing a PMO can be quite difficult, few studies discuss the specific challenges involved, and how organizations can overcome them. To address this gap in existing knowledge, we conducted a Delphi study to (1) identify the challenges of implementing a PMO for managing IT projects, (2) rank these challenges in order of importance, (3) discover ways in which some organizations have overcome the top-ranked challenges, and (4) understand the role of PMO structure, metrics, and tools in the implementation of a PMO.We identified 34 unique challenges to implementing a PMO and refined this list to 13 challenges that our Delphi panelists considered most important. The top-three challenges were (1) rigid corporate culture and failure to manage organizational resistance to change, (2) lack of experienced project managers (PMs) and PMO leadership, and (3) lack of appropriate change management strategy. Through follow-up interviews with selected panelists, we identified a series of actions that can be taken to overcome these challenges including having a strong PMO champion, starting small and demonstrating the value of the PMO, obtaining support from opinion leaders, hiring an experienced program manager who understands the organization, bringing the most talented PMs into the PMO implementation team, adopting a flexible change management strategy, and standardizing processes prior to PMO implementation. The interviews were also used to better understand the role of PMO structure, metrics, and tools. In terms of PMO structure, we found that ‘light’ PMOs were more likely to be implemented successfully. Most organizations eschew formal metrics, instead relying on subjective indicators of PMO success. Lastly, it appears that PMO tools are difficult to implement unless a project management culture has been established.

Wafer direct bonding refers to the process of adhesion of two flat mirror-polished wafers without using any intermediate gluing layers in ambient air or vacuum at room temperature. The adhesion of the two wafers occurs due to attractive long range van der Waals or hydrogen bonding forces. At room temperature the bonding energy of the interface is low and higher temperature annealing of the bonded wafer pairs has to be carried out to enhance the bonding energy. In this paper, we describe the prerequisites for the wafer-bonding process to occur and the methods to prepare the suitable surfaces for wafer bonding. The characterization techniques to assess the quality of the bonded interfaces and to measure the bonding energy are presented. Next, the applications of wafer direct bonding in the fabrication of novel engineered substrates such as "silicon-on-insulator" and other "on-insulator" substrates are detailed. These novel substrates, often called hybrid substrates, are fabricated using wafer bonding and layer splitting via a high dose hydrogen/helium implantation and subsequent annealing. The specifics of this process, also known as the smart-cut process, are introduced. Finally, the role of wafer bonding in future nanotechnology applications such as nanotransistor fabrication, three-dimensional integration for high-performance micro/nanoelectronics, nanotemplates based on twist bonding, and nano-electro-mechanical systems is discussed

Abstract A comparative study of urban mobile radio wave propagation characteristics in the UHF band has been carried out based on the field strength measurements conducted inside a moving train in two widely varying urban environments. Various prediction techniques are employed to compare the observed path losses with those of predicted ones in both the environments. The major fall out of the present study is the modification of Xia's model, which gave better agreement with the observed results. The deviations of the prediction techniques and their suitability have been discussed. Copyright © 2006 John Wiley & Sons, Ltd.