arXiv subjects

Ong Wee Hong

Publications and source records attributed to Ong Wee Hong.

3 recordsLinked to original sources

arXiv2023

Model Evaluation for Domain Identification of Unknown Classes in Open-World Recognition: A Proposal

Open-World Recognition (OWR) is an emerging field that makes a machine learning model competent in rejecting the unknowns, managing them, and incrementally adding novel samples to the base knowledge. However, this broad objective is not practical for an agent that works on a specific task. Not all rejected samples will be used for learning continually in the future. Some novel images in the open environment may not belong to the domain of interest. Hence, identifying the unknown in the domain of interest is essential for a machine learning model to learn merely the important samples. In this study, we propose an evaluation protocol for estimating a model's capability in separating unknown in-domain (ID) and unknown out-of-domain (OOD). We evaluated using three approaches with an unknown domain and demonstrated the possibility of identifying the domain of interest using the pre-trained parameters through traditional transfer learning, Automated Machine Learning (AutoML), and Nearest Class Mean (NCM) classifier with First Integer Neighbor Clustering Hierarchy (FINCH). We experimented with five different domains: garbage, food, dogs, plants, and birds. The results show that all approaches can be used as an initial baseline yielding a good accuracy. In addition, a Balanced Accuracy (BACCU) score from a pre-trained model indicates a tendency to excel in one or more domains of interest. We observed that MobileNetV3 yielded the highest BACCU score for the garbage domain and surpassed complex models such as the transformer network. Meanwhile, our results also suggest that a strong representation in the pre-trained model is important for identifying unknown classes in the same domain. This study could open the bridge toward open-world recognition in domain-specific tasks where the relevancy of the unknown classes is vital.

cs.CV↗

arXiv2023

Deep Reinforcement Learning-Based Mapless Crowd Navigation with Perceived Risk of the Moving Crowd for Mobile Robots

Current state-of-the-art crowd navigation approaches are mainly deep reinforcement learning (DRL)-based. However, DRL-based methods suffer from the issues of generalization and scalability. To overcome these challenges, we propose a method that includes a Collision Probability (CP) in the observation space to give the robot a sense of the level of danger of the moving crowd to help the robot navigate safely through crowds with unseen behaviors. We studied the effects of changing the number of moving obstacles to pay attention during navigation. During training, we generated local waypoints to increase the reward density and improve the learning efficiency of the system. Our approach was developed using deep reinforcement learning (DRL) and trained using the Gazebo simulator in a non-cooperative crowd environment with obstacles moving at randomized speeds and directions. We then evaluated our model on four different crowd-behavior scenarios. The results show that our method achieved a 100% success rate in all test settings. We compared our approach with a current state-of-the-art DRL-based approach, and our approach has performed significantly better, especially in terms of social safety. Importantly, our method can navigate in different crowd behaviors and requires no fine-tuning after being trained once. We further demonstrated the crowd navigation capability of our model in real-world tests.

cs.RO↗

arXiv2022

DSCOT: An NFT-Based Blockchain Architecture for the Authentication of IoT-Enabled Smart Devices in Smart Cities

Smart city architecture brings all the underlying architectures, i.e., Internet of Things (IoT), Cyber-Physical Systems (CPSs), Internet of Cyber-Physical Things (IoCPT), and Internet of Everything (IoE), together to work as a system under its umbrella. The goal of smart city architecture is to come up with a solution that may integrate all the real-time response applications. However, the cyber-physical space poses threats that can jeopardize the working of a smart city where all the data belonging to people, systems, and processes will be at risk. Various architectures based on centralized and distributed mechanisms support smart cities; however, the security concerns regarding traceability, scalability, security services, platform assistance, and resource management persist. In this paper, private blockchain-based architecture Decentralized Smart City of Things (DSCoT) is proposed. It actively utilizes fog computing for all the users and smart devices connected to a fog node in a particular management system in a smart city, i.e., a smart house or hospital, etc. Non-fungible tokens (NFTs) have been utilized for representation to define smart device attributes. NFTs in the proposed DSCoT architecture provide devices and user authentication (IoT) functionality. DSCoT has been designed to provide a smart city solution that ensures robust security features such as Confidentiality, Integrity, Availability (CIA), and authorization by defining new attributes and functions for Owner, User, Fog, and IoT devices authentication. The evaluation of the proposed functions and components in terms of Gas consumption and time complexity has shown promising results. Comparatively, the Gas consumption for minting DSCoT NFT showed approximately 27%, and a DSCoT approve() was approximately 11% more efficient than the PUF-based NFT solution.

cs.CR↗