| 1. |
- Zhang, Tianle, et al.
(author)
-
A Joint Deep Learning and Internet of Medical Things Driven Framework for Elderly Patients
- 2020
-
In: IEEE Access. - : IEEE. - 2169-3536. ; 8:2020, s. 75822-75832
-
Journal article (peer-reviewed)abstract
- Deep learning (DL) driven cardiac image processing methods manage and monitor the massive medical data collected by the internet of things (IoT) based on wearable devices. A Joint DL and IoT platform are known as Deep-IoMT that extracts the accurate cardiac image data from noisy conventional devices and tools. Besides, smart and dynamic technological trends have caught the attention of every corner such as, healthcare, which is possible through portable and lightweight sensor-enabled devices. Tiny size and resource-constrained nature restrict them to perform several tasks at a time. Thus, energy drain, limited battery lifetime, and high packet loss ratio (PLR) are the keys challenges to be tackled carefully for ubiquitous medical care. Sustainability (i.e., longer battery lifetime), energy efficiency, and reliability are the vital ingredients for wearable devices to empower a cost-effective and pervasive healthcare environment. Thus, the key contribution of this paper is the sixth fold. First, a novel self-adaptive power control-based enhanced efficient-aware approach (EEA) is proposed to reduce energy consumption and enhance the battery lifetime and reliability. The proposed EEA and conventional constant TPC are evaluated by adopting real-time data traces of static (i.e., sitting) and dynamic (i.e., cycling) activities and cardiac images. Second, a novel joint DL-IoMT framework is proposed for the cardiac image processing of remote elderly patients. Third, DL driven layered architecture for IoMT is proposed. Forth, the battery model for IoMT is proposed by adopting the features of a wireless channel and body postures. Fifth, network performance is optimized by introducing sustainability, energy drain, and PLR and average threshold RSSI indicators. Sixth, a Use-case for cardiac image-enabled elderly patient's monitoring is proposed. Finally, it is revealed through experimental results in MATLAB that the proposed EEA scheme performs better than the constant TPC by enhancing energy efficiency, sustainability, and reliability during data transmission for elderly healthcare.
|
|
| 2. |
- Butun, Ismail, et al.
(author)
-
AI-Enabled Smart Sensing Technologies for Human-Centered Healthcare Applications
- 2022
-
In: Sensors. - : MDPI Multidisciplinary Digital Publishing Institute. - 1424-8220. ; , s. 2
-
Journal article (pop. science, debate, etc.)abstract
- Lead Guest Editor, Special Issue ‘AI-Enabled Smart Sensing Technologies for Human-Centered Healthcare Applications’, Sensors, MDPI, from Jan 2022 to Oct 2023, https://www.mdpi.com/journal/sensors/special_issues/AESSTHCHA
|
|
| 3. |
- Khattak, Muhammad Uzair, et al.
(author)
-
Self-regulating Prompts: Foundational Model Adaptation without Forgetting
- 2023
-
In: 2023 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2023). - : IEEE COMPUTER SOC. - 9798350307184 - 9798350307191 ; , s. 15144-15154
-
Conference paper (peer-reviewed)abstract
- Prompt learning has emerged as an efficient alternative for fine-tuning foundational models, such as CLIP, for various downstream tasks. Conventionally trained using the task-specific objective, i.e., cross-entropy loss, prompts tend to overfit downstream data distributions and find it challenging to capture task-agnostic general features from the frozen CLIP. This leads to the loss of the model's original generalization capability. To address this issue, our work introduces a self-regularization framework for prompting called PromptSRC (Prompting with Self-regulating Constraints). PromptSRC guides the prompts to optimize for both task-specific and task-agnostic general representations using a three-pronged approach by: (a) regulating prompted representations via mutual agreement maximization with the frozen model, (b) regulating with selfensemble of prompts over the training trajectory to encode their complementary strengths, and (c) regulating with textual diversity to mitigate sample diversity imbalance with the visual branch. To the best of our knowledge, this is the first regularization framework for prompt learning that avoids overfitting by jointly attending to pre-trained model features, the training trajectory during prompting, and the textual diversity. PromptSRC explicitly steers the prompts to learn a representation space that maximizes performance on downstream tasks without compromising CLIP generalization. We perform extensive experiments on 4 benchmarks where PromptSRC overall performs favorably well compared to the existing methods. Our code and pre-trained models are publicly available at: https://github.com/muzairkhattak/PromptSRC.
|
|
| 4. |
- Muzammal, Muhammad, et al.
(author)
-
A Multi-sensor Data Fusion Enabled Ensemble Approach for Medical Data from Body Sensor Networks
- 2020
-
In: Information Fusion. - : Elsevier. - 1566-2535 .- 1872-6305. ; 53:2020, s. 155-164
-
Journal article (peer-reviewed)abstract
- Wireless Body Sensor Network (BSNs) are wearable sensors with varying sensing, storage, computation, and transmission capabilities. When data is obtained from multiple devices, multi-sensor fusion is desirable to transform potentially erroneous sensor data into high quality fused data. In this work, a data fusion enabled Ensemble approach is proposed to work with medical data obtained from BSNs in a fog computing environment. Daily activity data is obtained from a collection of sensors which is fused together to generate high quality activity data. The fused data is later input to an Ensemble classifier for early heart disease prediction. The ensembles are hosted in a Fog computing environment and the prediction computations are performed in a decentralised manners. The results from the individual nodes in the fog computing environment are then combined to produce a unified output. For the classification purpose, a novel kernel random forest ensemble is used that produces significantly better quality results than random forest. An extensive experimental study supports the applicability of the solution and the obtained results are promising, as we obtain 98% accuracy when the tree depth is equal to 15, number of estimators is 40, and 8 features are considered for the prediction task.
|
|
| 5. |
- Naseer, Muzammal, et al.
(author)
-
Cross-Domain Transferability of Adversarial Perturbations
- 2019
-
In: ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 32 (NIPS 2019). - : NEURAL INFORMATION PROCESSING SYSTEMS (NIPS).
-
Conference paper (peer-reviewed)abstract
- Adversarial examples reveal the blind spots of deep neural networks (DNNs) and represent a major concern for security-critical applications. The transferability of adversarial examples makes real-world attacks possible in black-box settings, where the attacker is forbidden to access the internal parameters of the model. The underlying assumption in most adversary generation methods, whether learning an instance-specific or an instance-agnostic perturbation, is the direct or indirect reliance on the original domain-specific data distribution. In this work, for the first time, we demonstrate the existence of domain-invariant adversaries, thereby showing common adversarial space among different datasets and models. To this end, we propose a framework capable of launching highly transferable attacks that crafts adversarial patterns to mislead networks trained on entirely different domains. For instance, an adversarial function learned on Paintings, Cartoons or Medical images can successfully perturb ImageNet samples to fool the classifier, with success rates as high as similar to 99% (l(infinity) <= 10). The core of our proposed adversarial function is a generative network that is trained using a relativistic supervisory signal that enables domain-invariant perturbations. Our approach sets the new state-of-the-art for fooling rates, both under the white-box and black-box scenarios. Furthermore, despite being an instance-agnostic perturbation function, our attack outperforms the conventionally much stronger instance-specific attack methods.
|
|
| 6. |
- Sodhro, Ali Hassan, et al.
(author)
-
An adaptive QoS computation for medical data processing in intelligent healthcare applications
- 2020
-
In: Neural Computing and Applications. - : Springer. - 0941-0643 .- 1433-3058. ; 32:2020, s. 723-734
-
Journal article (peer-reviewed)abstract
- Efficient computation of quality of service (QoS) during medical data processing through intelligent measurement methods is one of the mandatory requirements of the medial healthcare world. However, emergency medical services often involve transmission of critical data, thus having stringent requirements for network quality of service (QoS). This paper contributes in three distinct ways. First, it proposes the novel adaptive QoS computation algorithm (AQCA) for fair and efficient monitoring of the performance indicators, i.e., transmission power, duty cycle and route selection during medical data processing in healthcare applications. Second, framework of QoS computation in medical applications is proposed at physical, medium access control (MAC) and network layers. Third, QoS computation mechanism with proposed AQCA and quality of experience (QoE) is developed. Besides, proper examination of QoS computation for medical healthcare application is evaluated with 4–10 inches large-screen user terminal (UT) devices (for example, LCD panel size, resolution, etc.). These devices are based on high visualization, battery lifetime and power optimization for ECG service in emergency condition. These UT devices are used to achieve highest level of satisfaction in terms, i.e., less power drain, extended battery lifetime and optimal route selection. QoS parameters with estimation of QoE perception identify the degree of influence of each QoS parameters on the medical data processing is analyzed. The experimental results indicate that QoS is computed at physical, MAC and network layers with transmission power (− 15 dBm), delay (100 ms), jitter (40 ms), throughput (200 Bytes), duty cycle (10%) and route selection (optimal). Thus it can be said that proposed AQCA is the potential candidate for QoS computation than Baseline for medical healthcare applications.
|
|
| 7. |
- Sodhro, Ali Hassan, et al.
(author)
-
Towards 5G-Enabled Self Adaptive Green and Reliable Communication in Intelligent Transportation System
- 2021
-
In: IEEE transactions on intelligent transportation systems (Print). - : IEEE. - 1524-9050 .- 1558-0016. ; 22:8, s. 5223-5231
-
Journal article (peer-reviewed)abstract
- Fifth generation (5G) technologies have become the center of attention in managing and monitoring high-speed transportation system effectively with the intelligent and self-adaptive sensing capabilities. Besides, the boom in portable devices has witnessed a huge breakthrough in the data driven vehicular platform. However, sensor-based Internet of Things (IoT) devices are playing the major role as edge nodes in the intelligent transportation system (ITS). Thus, due to high mobility/speed of vehicles and resource-constrained nature of edge nodes more data packets will be lost with high power drain and shorter battery life. Thus, this research significantly contributes in three ways. First, 5G-based self-adaptive green (i.e., energy efficient) algorithm is proposed. Second, a novel 5G-driven reliable algorithm is proposed. Proposed joint energy efficient and reliable approach contains four layers, i.e., application, physical, networks, and medium access control. Third, a novel joint energy efficient and reliable framework is proposed for ITS. Moreover, the energy and reliability in terms of received signal strength (RSSI) and hence packet loss ratio (PLR) optimization is performed under the constraint that all transmitted packets must utilize minimum transmission power with high reliability under particular active time slot. Experimental results reveal that the proposed approach (with Cross Layer) significantly obtains the green (55%) and reliable (41%) ITS platform unlike the Baseline (without Cross Layer) for aging society.
|
|
| 8. |
- Sodhro, Ali Hassan, et al.
(author)
-
Towards Blockchain-Enabled Security Technique for Industrial Internet of Things Based Decentralized Applications
- 2020
-
In: Journal of Grid Computing. - Springer : Springer. - 1570-7873 .- 1572-9184. ; 18:2020, s. 615-628
-
Journal article (peer-reviewed)abstract
- As the Industrial Internet of Things (IIoT) is one of the emerging trends and paradigm shifts to revolutionize the traditional industries with the fourth wave of evolution or transform it into Industry 4.0. This all is merely possible with the sensor-enabled technologies, e.g., wireless sensor networks (WSNs) in various landscapes, where security provisioning is one of the significant challenges for miniaturized power hungry networks. Due to the increasing demand for the commercial Internet of things (IoT) devices, smart devices are also extensively adopted in industrial applications. If these devices are compromising the date/information, then there will be a considerable loss and critical issues, unlike information compromising level by the commercial IoT devices. So emerging industrial processes and smart IoT based methods in medical industries with state-of-the-art blockchain security techniques have motivated the role of secure industrial IoT. Also, frequent changes in android technology have increased the security of the blockchain-based IIoT system management. It is very vital to develop a novel blockchain-enabled cyber-security framework and algorithm for industrial IoT by adopting random initial and master key generation mechanisms over long-range low-power wireless networks for fast encrypted data processing and transmission. So, this paper has three remarkable contributions. First, a blockchain-driven secure, efficient, reliable, and sustainable algorithm is proposed. It can be said that the proposed solution manages keys randomly by introducing the chain of blocks with less power drain, a small number of cores, will slightly more communication and computation bits. Second, an analytic hierarchy process (AHP) based intelligent decision-making approach for the secure, concurrent, interoperable, sustainable, and reliable blockchain-driven IIoT system. AHP based solution helps the industry experts to select the more relevant and critical parameters such as (reliability in-line with a packet loss ratio), (convergence in mapping with delay), and (interoperability in association with throughput) for improving the yield of the product in the industry. Third, sustainable technology-oriented services are supporting to propose the novel cloud-enabled framework for the IIoT platform for regular monitoring of the products in the industry. Moreover, experimental results reveal that proposed approach is a potential candidate for the blockchain-driven IIoT system in terms of reliability, convergence, and interoperability with a strong foundation to predict the techniques and tools for the regulation of the adaptive system from Industry 4.0 aspect.
|
|
| 9. |
- Talat, Romana, et al.
(author)
-
A decentralised approach to privacy preserving trajectory mining
- 2020
-
In: Future Generation Computer Application. - : Elsevier. - 0167-739X .- 1872-7115. ; 102:2020, s. 382-392
-
Journal article (peer-reviewed)abstract
- Large volumes of mobility data is collected in various application domains. Enterprise applications are designed on the notion of centralised data control where the proprietary of the data rests with the enterprise and not with the user. This has consequences as evident by the occasional privacy breaches. Trajectory mining is an important data mining problem, however, trajectory data can disclose sensitive location information about users. In this work, we propose a decentralised blockchain-enabled privacy-preserving trajectory data mining framework where the proprietary of the data rests with the user and not with the enterprise. We formalise the privacy preservation in trajectory data mining settings, present a proposal for privacy preservation, and implement the solution as a proof-of-concept. A comprehensive experimental evaluation is conducted to assess the applicability of the system. The results show that the proposed system yields promising results for blockchain-enabled privacy preservation in user trajectory data.
|
|
| 10. |
- Wasim, Syed Talal, et al.
(author)
-
Video-FocalNets: Spatio-Temporal Focal Modulation for Video Action Recognition
- 2023
-
In: 2023 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV 2023). - : IEEE COMPUTER SOC. - 9798350307184 - 9798350307191 ; , s. 13732-13743
-
Conference paper (peer-reviewed)abstract
- Recent video recognition models utilize Transformer models for long-range spatio-temporal context modeling. Video transformer designs are based on self-attention that can model global context at a high computational cost. In comparison, convolutional designs for videos offer an efficient alternative but lack long-range dependency modeling. Towards achieving the best of both designs, this work proposes Video-FocalNet, an effective and efficient architecture for video recognition that models both local and global contexts. Video-FocalNet is based on a spatio-temporal focal modulation architecture that reverses the interaction and aggregation steps of self-attention for better efficiency. Further, the aggregation step and the interaction step are both implemented using efficient convolution and element-wise multiplication operations that are computationally less expensive than their self-attention counterparts on video representations. We extensively explore the design space of focal modulation-based spatio-temporal context modeling and demonstrate our parallel spatial and temporal encoding design to be the optimal choice. Video-FocalNets perform favorably well against the state-of-the-art transformer-based models for video recognition on five large-scale datasets (Kinetics-400, Kinetics-600, SS-v2, Diving-48, and ActivityNet-1.3) at a lower computational cost. Our code/models are released at https://github.com/TalalWasim/Video-FocalNets.
|
|
