Search

The interpretability of deep neural networks has become a subject of great interest within the medical and healthcare domain. This attention stems from concerns regarding transparency, legal and ethical considerations, and the medical significance of predictions generated by these deep neural networks in clinical decision support systems. To address this matter, our study delves into the application of four well-established interpretability methods: Local Interpretable Model-agnostic Explanations (LIME), Shapley Additive exPlanations (SHAP), Gradient-weighted Class Activation Mapping (Grad-CAM), and Layer-wise Relevance Propagation (LRP). Leveraging the approach of transfer learning with a multi-label-multi-class chest radiography dataset, we aim to interpret predictions pertaining to specific pathology classes. Our analysis encompasses both single-label and multi-label predictions, providing a comprehensive and unbiased assessment through quantitative and qualitative investigations, which are compared against human expert annotation. Notably, Grad-CAM demonstrates the most favorable performance in quantitative evaluation, while the LIME heatmap score segmentation visualization exhibits the highest level of medical significance. Our research underscores both the outcomes and the challenges faced in the holistic approach adopted for assessing these interpretability methods and suggests that a multimodal-based approach, incorporating diverse sources of information beyond chest radiography images, could offer additional insights for enhancing interpretability in the medical domain., Comment: This version contains extensive modifications compared to the previous version. The published version of this article can be obtained using the following link: https://doi.org/10.5617/nmi.10471 Code Repository: https://github.com/anondo1969/SHAMSUL

The Internet of Things(IoT) will revolutionize the Future Internet through ubiquitous sensing. One of the challenges of having the hundreds of billions of devices that are estimated to be deployed would be rise of an enormous amount of data, along with the devices ability to manage. This paper presents an approach as a controller solution and designed specifically for autonomous management, connectivity and data interoperability in an IoT gateway. The approach supports distributed IoT nodes with both management and data interoperability with other cloud-based solutions. The concept further allows gateways to easily collect and process interoperability of data from IoT devices. We demonstrated the feasibility of the approach and evaluate its advantages regarding deep sensing and autonomous enabled gateway as an edge computational intelligence.

There is increasing demand for control of multi-robot and as well distributing large amounts of content to cluster of Unmanned Aerial Vehicles (UAV) on the operation. In recent years several large-scale accidents have happened. To facilitate rescue operations and gather information, the technology that can access and map inaccessible areas is needed. This paper presents a disruptive approach to address the issues with communication, data collection and data sharing for UAV units in inaccessible or dead zones and We demonstrated feasibility of the approach and evaluate its advantages over the Ad Hoc architecture involving autonomous gateways

With the rise of the Internet of Things (IoT), this means recognizing the need for architectures to handle billions of devices and their interactions. A virtual world engine at the massively multiplayer scale is a massively multiplayer online world (MMOW); one thing virtual world engines realized when going into the scale of MMOs, is the cost of maintaining a potentially quadratic number of interactions between a massive number of objects, laid out in a spatial dimension. Research into IoT was fueled by research in wireless sensor networks, but rather than start from a device perspective, this article looks at how architectures deal with interacting entities at large scale. The domain of MMOWs is examined for properties that are affected by scale. Thereafter the domain of IoT is evaluated to see if each of those properties are found and how each is handled. By comparing the current state of the art of MMOWs and IoT, with respect to scalability, the problem of scaling IoT is explicated, as well as the problem of incorporating an MMOW with IoT into a pervasive platform. Three case studies of a MMOW interfacing with IoT are presented in closing., Comment: 10 pages, 2 figures

A novel conceptual framework is presented in this paper with an aim to standardize and virtualize Internet of Things(IoT) infrastructure through deploying OpenFlow technology. The framework can receivee services based on context information leaving the current infrastructure unchanged. This framework allows the active collaboration of heterogeneous devices and protocols. Moreover it is capable to model placement of physical objects, manage the system and to collect information for services deployed on an IoT infrastructure. Our proposed IoT virtualization is applicable to a random topology scenario which makes it possible to 1) share flow sensors resources 2) establish multioperational sensor networks, and 3) extend reachability within the framework without establishing any further physical networks. Flow sensors achieve better results comparable to the typical sensors with respect to packet generation, reachability, simulation time, throughput, energy consumption point of view. Even better results are possible through utilizing multicast groups in large scale networks., Comment: 12 pages, 19 Figures, 1 table

In state-of-the-art Pervasive Computing, it is envisioned that unlimited access to information will be facilitated for anyone and anything. Wireless sensor networks will play a pivotal role in the stated vision. This reflects the phenomena where any situation can be sensed and analyzed anywhere. It makes heterogeneous context ubiquitous. Clustering context is one of the techniques to manage ubiquitous context information efficiently to maximize its potential. Logical-clustering is useful to share real-time context where sensors are physically distributed but logically clustered. This paper investigates the network performance of logical-clustering based on ns-3 simulations. In particular reliability, scalability, and reachability in terms of delay, jitter, and packet loss for the logically clustered network have been investigated. The performance study shows that jitter demonstrates 40 % and 44 % fluctuation for 200 % increase in the node per cluster and 100 % increase in the cluster size respectively. Packet loss exhibits only 18 % increase for 83 % increase in the packet flow-rate., Comment: 13 pages, 15 Figures and published on International Journal of Computer Science Issues, IJCSI Volume 10 Issue 5, September 2013