No abstract

No abstract

Transfer Learning (TL) plays a crucial role when a given dataset has insufficient labeled examples to train an accurate model. In such scenarios, the knowledge accumulated within a model pre-trained on a source dataset can be transferred to a target dataset, resulting in the improvement of the target model. Though TL is found to be successful in the realm of imagebased applications, its impact and practical use in Natural Language Processing (NLP) applications is still a subject of research. Due to their hierarchical architecture, Deep Neural Networks (DNN) provide flexibility and customization in adjusting their parameters and depth of layers, thereby forming an apt area for exploiting the use of TL. In this paper, we report the results and conclusions obtained from extensive empirical experiments using a Convolutional Neural Network (CNN) and try to uncover thumb rules to ensure a successful positive transfer. In addition, we also highlight the flawed means that could lead to a negative transfer. We explore the transferability of various layers and describe the effect of varying hyper-parameters on the transfer performance. Also, we present a comparison of accuracy value and model size against state-of-the-art methods. Finally, we derive inferences from the empirical results and provide best practices to achieve a successful positive transfer.

Abstract:The Raspberry Pi and its variants have brought with them an aura of change in the world of embedded systems. With their impressive computation and communication capabilities and low footprint, these devices have thrown open the possibility of realizing a network of things in a very cost-effective manner. While such networks offer good solutions to prominent issues, they are indeed a long way from being smart or intelligent. Most of the currently available implementations of such a network of devices involve a centralized cloud-based server that contributes to making the necessary intelligent decisions, leaving these devices fairly underutilized. Though this paradigm provides for an easy and rapid solution, they have limited scalability, are less robust and at times prove to be expensive. In this paper, we introduce the concept of Agents on Raspberry Pi (AgPi) as a cyber solution to enhance the smartness and flexibility of such embedded networks of physical devices in a decentralized manner. The use of a Multi-Agent System (MAS) running on Raspberry Pis aids agents, both static and mobile, to govern the various activities within the network. Agents can act autonomously or on behalf of a human user and can collaborate, learn, adapt and act, thus contributing to embedded intelligence. This paper describes how Tartarus, a multi-agent platform, embedded on Raspberry Pis that constitute a network, can bring the best out of the system. To reveal the versatility of the concept of AgPi, an application for a Location-Aware and Tracking Service (LATS) is presented. The results obtained from a comparison of data transfer cost between the conventional cloud-based approach with AgPi have also been included.

No abstract

No abstract

No abstract

No abstract