There has been an upsurge recently in investigating machine learning techniques for activity recognition (AR) problems as they have very effective extracting and knowledge from the datasets. The technique ranges heuristically derived hand‐crafted feature‐based traditional algorithms to developed hierarchically self‐evolving deep algorithms. AR continues remain a challenging problem uncontrolled smart environments despite amount of work contributed by researcher this field. complex, volatile,...

We investigate the problem of making human activity recognition (AR) scalable-i.e., allowing AR classifiers trained in one context to be readily adapted a different contextual domain. This is important because technologies can achieve high accuracy if are for specific individual or device, but show significant degradation when same classifier applied context-e.g., device located at on-body position. To allow such adaptation without requiring onerous step collecting large volumes labeled...

Abstract The rapid and impromptu interest in the coupling of machine learning (ML) algorithms with wearable contactless sensors aimed at tackling real‐world problems warrants a pedagogical study to understand all aspects this research direction. Considering aspect, survey aims review state‐of‐the‐art literature on ML algorithms, methodologies, hypotheses adopted solve challenges domain sports. First, we categorize into three main fields: , computer vision wireless mobile‐based applications ....

Activity recognition in smart environment has been investigated rigorously recent years. Researchers are enhancing the underlying activity discovery and process by adding various dimensions functionalities. But one significant barrier still persists which is collecting ground truth information. Ground very important to initialize a supervised learning of activities. Due large variety number Activities Daily Living (ADLs), acknowledging them way non-trivial research problem. Most previous...

A smart home aims at building intelligence automation with a goal to provide its inhabitants maximum possible comfort, minimize the resource consumption and thus overall cost of maintaining home. 'Context awareness' is perhaps most salient feature such an intelligent environment. Clearly, inhabitant's mobility activities play significant role in defining his contexts around Although there exists optimal algorithm for location activity tracking single inhabitant, correlation dependence...

Deep learning architectures have been applied increasingly in multi-modal problems which has empowered a large number of application domains needing much less human supervision the process. As unlabeled data are abundant most domains, deep getting popular to extract meaningful information out these volume data. One major caveat is that training phase demands both computational time and system resources higher than shallow algorithms it posing difficult challenge for researchers implement...

Summary form only given. This article realizes the vision of mobile grid computing by proposing a fair pricing strategy and an optimal, static job allocation scheme. Mobile devices has not yet been integrated into platforms mainly due to their inherent limitations in processing storage capacity, power bandwidth shortages. However, millions laptops, PDAs other remain unused most time this huge resource repository can be potentially utilized environment. Here, we propose game theoretic model,...

We propose a hybrid approach for recognizing complex Activities of Daily Living that lie between the two extremes intensive use body-worn sensors and infrastructural sensors. Our harnesses power (e.g., motion sensors) to provide additional `hidden' context room-level location) an individual combines this with smartphone-based sensing micro-level postural/locomotive states. The major novelty is our focus on multi-inhabitant environments, where we show how spatiotemporal constraints can be...

Dementia is a clinical syndrome of cognitive deficits that involves both memory and functional impairments. While disruptions in cognition striking feature dementia, it also closely coupled with changes behavioral health older adults. In this paper, we investigate the challenges improving automatic assessment by better exploiting emerging physiological sensors conjunction ambient real field environment IRB approval. We hypothesize individuals can be estimated tracking their daily activities...

In pervasive computing environments, understanding the context of an entity is essential for adapting application behavior to changing situations. our view, a high-level representation user or entity's state and can capture location, activities, social relationships, capabilities, etc. Inherently, however, these metrics are difficult using uni-modal sensors only, must therefore be inferred with help multi-modal sensors. However key challenge in supporting context-aware how determine...

Machine learning models are bounded by the credibility of ground truth data used for both training and testing. Regardless problem domain, this annotation is objectively manual tedious as it needs considerable amount human intervention. With advent Active Learning with multiple annotators, burden can be somewhat mitigated actively acquiring labels most informative instances. However, annotators varying degrees expertise poses new set challenges in terms quality label received availability...

Cybercrime is a growing threat to organizations and individuals worldwide, with criminals using sophisticated techniques breach security systems steal sensitive data. This paper aims comprehensively survey the latest advancements in cybercrime prediction, highlighting relevant research. For this purpose, we reviewed more than 150 research articles discussed 50 most recent appropriate ones. We start review some standard methods cybercriminals use then focus on machine deep learning...

Recent advancements in deep learning-based wearable human action recognition (wHAR) have improved the capture and classification of complex motions, but adoption remains limited due to lack expert annotations domain discrepancies from user variations. Limited hinder model's ability generalize out-of-distribution samples. While data augmentation can improve generalizability, unsupervised techniques must be applied carefully avoid introducing noise. Unsupervised adaptation (UDA) addresses by...

Natural language is a flexible and intuitive modality for conveying directions commands to robot but presents number of computational challenges. Diverse words phrases must be mapped into structures that the can understand, elements in those grounded an uncertain environment. In this paper we present micro-air vehicle (MAV) capable following natural through previously labeled We extend our previous work understanding 2D three dimensions, accommodating new verb modifiers such as go up down,...