Chromosome straightening plays an important role in karyotype analysis. Common straightening methods usually adopt geometric algorithms, which tend to affect the chromosome banding patterns in the process of straightening, resulting in feature changes, loss of details, and poor generalization. To solve these problems, this paper proposes a novel straightening method based on disentanglement representation learning. Our method consists of two main components: the Disentanglement Representation Encoder (DRE) and the Straightening Generator (SG), where DRE discovers and disentangles the bent representation and the content representation in the latent space, while SG is used to generate the straightened chromosome images based on the disentangled representations. Leveraging the bent representation and the content representation disentangled by DRE, our method produces the straightened representation by reducing the bent representation while keeping the content representation unchanged, making straightening chromosome without changing its banding patterns possible. Evaluation results on both the Frechet Initiation Distance (FID) and the Downstream Classification Accuracy (DCA) metrics show that our method achieves good performance.

Chromosome instance segmentation plays a crucial role in chromosomal karyotype analysis. However, the overlapping of chromosome instances and their individual morphological differences make accurate chromosome instance segmentation a challenging task. Especially in handling overlapping chromosome instances, traditional segmentation methods tend to confuse instances with one another. To solve these problems, this paper proposes an innovative method named Adaptive2Former. It builds upon our novel devised Adaptive Query Decoder (AQD) module to enhance segmentation precision. The AQD effectively utilizes the [cls] token from the backbone network to dynamically generate adaptive query vectors instead of using fixed queries. This new design leverages the semantic information inside the input images, producing representations more conducive to subsequent segmentation module, thereby improving the model’s segmentation performance. Experiments conducted on our dataset demonstrate that the proposed Adaptive2Former significantly enhances the performance of chromosome instance segmentation compared to Mask2Former and other existing models, achieving results of 97.65% mAP75 and 0.21 Dice Loss.

Human chromosomes carry genetic information about our life. Chromosome classification is crucial for karyotype analysis. Existing chromosome classification methods do not take into account reasoning, such as: analyzing the relationship between variables, modeling uncertainty, and performing causal reasoning. In this paper, we introduce a knowledge engine for reasoning-based human chromosome classification that stores knowledge of chromosomes via a novel representation structure, the Chromosome Part Description (CPD), and reasons over CPDs by utilizing the probability tree model (PTM) for classification. Each CPD keeps information on a particular feature of chromosomes, while the PTM provides causal reasoning capability taking CPDs as nodes and dependencies between CPDs and types as edges. Experimental results show that the proposed knowledge engine’s performance increases when providing more CPDs and achieves 100% classification accuracy with more than three CPDs.

Karyotyping requires chromosome instances to be segmented and classified from the metaphase images. One of the difficulties in chromosome segmentation is that the chromosomes are randomly positioned in the image, and there is a great chance for chromosomes to be touched or overlap with others. It is always much easier for operators and automatic programs to tackle images without overlapping chromosomes than ones with largely overlapped chromosomes. In order to reduce the processing difficulty, adding a smart image selection procedure ahead of segmentation is practical and necessary. In this paper, we introduce the Smart Karyotyping Image Selection (SKIS) based on Commonsense Knowledge Reasoning. The initial experiment demonstrates that the proposed approach can select the expected images based on reasoning and benefit following karyotyping processes.

This article proposes a mask refinement method for chromosome instance segmentation. The proposed method exploits the knowledge representation capability of Neural Knowledge DNA (NK-DNA) to capture the semantics of the chromosome’s shape, texture, and key points, and then it uses the captured knowledge to improve the accuracy and smoothness of the masks. We validate the method’s effectiveness on our latest high-resolution chromosome image dataset. The experimental results show that our proposed method’s mask average precision (MaskAP) is 3.66% higher than Mask R-CNN and outperforms advanced Cascade Mask R-CNN by 1.35%.

Over the years, several deep learning algorithms have been proposed for electroencephalography (EEG) signal classification. The performance of any learning method usually relies on the quality of the learned representation that provides semantic information for downstream tasks such as classification. Thus, it is crucial to improve the model’s representation learning capability. This paper proposes an Efficient Temporal Network with dual self-distillation for EEG signal classification, ETNEEG. It enhances the model’s representation learning by promoting mutual learning between higher-level and lower-level semantic information. The proposed ETNEEG consists of two main components: a parallel dual-network-based feature extractor called MLN-GRN and a dual self-distillation module. MLN-GRN includes a multi-scale local network (MLN) and a global relation network (GRN). MLN pays attention to local features of EEG data, and GRN is designed for learning global patterns of EEG data. Meanwhile, the dual self-distillation module extracts semantic information by mutual learning among the output layer and the low-level features. To evaluate the proposed method’s performance, seven widely used public EEG datasets, i.e., FaceDetection, FingerMovements, HandMovementDirection, MotorImagery, PenDigits, SelfRegulationSCP1, and SelfRegulationSCP2, are applied to a set of experiments. Experimental results demonstrate that the proposed ETNEEG achieves excellent performance on these datasets compared with fourteen existing algorithms.

In this paper, we propose a knowledge sharing oriented approach to enable a robot to reuse other robots' knowledge by adapting itself to the inverse dynamics model of the knowledge-sharing robot. The purpose of this work is to remove the heavy fine-tuning procedure required before using a new robot for a task via reusing other robots' knowledge. We use the Neural Knowledge DNA (NK-DNA) to help robots gain empirical knowledge and introduce a Knowledge Adaption Module (KAM) utilizing the deep neural networks (DNN) for knowledge reuse. The initial experiment shows that the target robot can adapt to the inverse dynamic model of the source robot via our KAM and reuse the knowledge shared by the source robot.

This paper proposes a novel approach that adds the interpretability to Neural Knowledge DNA (NK-DNA) via generating a decision tree. The NK-DNA is a promising knowledge representation approach for acquiring, storing, sharing, and reusing knowledge among machines and computing systems. We introduce the decision tree-based generative method for knowledge extraction and representation to make the NK-DNA more explainable. We examine our approach through an initial case study. The experiment results show that the proposed method can transform the implicit knowledge stored in the NK-DNA into explicitly represented decision trees bringing fair interpretability to neural network-based intelligent systems.

In this paper we propose a novel recommendation approach using past news click data and the Neural Knowledge DNA (NK-DNA). The Neural Knowledge DNA is a novel knowledge representation method designed to support discovering, storing, reusing, improving, and sharing knowledge among machines and computing systems. We examine our approach for news recommendation tasks on the MIND benchmark dataset. By taking advantages of NK-DNA, deep learning, and the SOEKS technologies, our approach can learn from users’ past behaviors to form reading preference of the user, and reuse learned knowledge for improving the recommendation performance.

The Internet of Things (IoT) has gained significant attention from industry as well as academia during the past decade. Smartness, however, remains a substantial challenge for IoT applications. Recent advances in networked sensor technologies, computing, and machine learning have made it possible for building new smart IoT applications. In this paper, we propose a novel approach: the Neural Knowledge DNA based Smart Internet of Things that enables IoT to extract knowledge from past experiences, as well as to store, evolve, share, and reuse such knowledge aiming for smart functions. By catching decision events, this approach helps IoT gather its own daily operation experiences, and it uses such experiences for knowledge discovery with the support of machine learning technologies. An initial case study is presented at the end of this paper to demonstrate how this approach can help IoT applications become smart: the proposed approach is applied to fitness wristbands to enable human action recognition.

With the rapid progress of information technologies, cars have been made increasingly intelligent. This allows cars to act as cognitive agents, i.e., to acquire knowledge and understanding of the driving habits and behavioral characteristics of drivers (i.e., driving behavioral fingerprint) through experience. Such knowledge can be then reused to facilitate the interaction between a car and its driver, and to develop better and safer car controls. In this paper, we propose a novel approach to extract the driver’s driving behavioral fingerprints based on our conceptual framework Experience-Oriented Intelligent Things (EOIT). EOIT is a learning system that has the potential to enable Internet of Cognitive Things (IoCT) where knowledge can be extracted from experience, stored, evolved, shared, and reused aiming for cognition and thus intelligent functionality of things. By catching driving data, this approach helps cars to collect the driver’s pedal and steering operations and store them as experience; eventually, it uses obtained experience for the driver’s driving behavioral fingerprint extraction. The initial experimental implementation is presented in the paper to demonstrate our idea, and the test results show that it outperforms the Deep Learning approaches (i.e., deep fully connected neural networks and recurrent neural networks/Long Short-Term Memory networks).

Together with the fast advancement of the Internet of Things (IoT), smart healthcare applications and systems are equipped with increasingly more wearable sensors and mobile devices. These sensors are used not only to collect data but also, and more importantly, to assist in daily activity tracking and analyzing of their users. Various human activity recognition (HAR) approaches are used to enhance such tracking. Most of the existing HAR methods depend on exploratory case-based shallow feature learning architectures, which struggle with correct activity recognition when put into real-life practice. To tackle this problem, we propose a novel approach that utilizes the convolutional neural networks (CNNs) and the attention mechanism for HAR. In the presented method, the activity recognition accuracy is improved by incorporating attention into multihead CNNs for better feature extraction and selection. Proof of concept experiments are conducted on a publicly available data set from wireless sensor data mining (WISDM) lab. The results demonstrate a higher accuracy of our proposed approach in comparison with the current methods.

In this article, we introduce a novel security mechanism, the Smart Synergy Security (3S). The mechanism uses the Set of Experience Knowledge Structure (SOEKS) and the synergy of security methods in different domains to provide the global optimal security strategy. The proposed strategy is taking into account the characteristics of information security (i.e. confidentiality, integrity, availability, controllability, and reviewability) imposed in different domains in Internet of Vehicles (IoV). The SOEKS is used to represent knowledge, and is combined with the data flow in each domain. Initial experiments demonstrate that the proposed approach is able to find the optimal solution under different conditions for multi-domain security problems in IoV.

In this paper, we propose a novel intrusion detection approach using past driving experience and the neural knowledge DNA for in-vehicle information system security. The neural knowledge DNA is a novel knowledge representation method designed to support discovering, storing, reusing, improving, and sharing knowledge among machines and computing systems. We examine our approach for classifying malicious vehicle control commands based on learning from past valid driving behavior data on a simulator.

The Internet of Things (IoT) has gained significant attention from industry as well as academia during the past decade.The main reason behind this interest is the capabilities of the IoT for seamlessly integrating classical networks and networked objects, and hence allowing people to create an intelligent environment based on this powerful integration. However, how to extract useful information from data produced by IoT and facilitate standard knowledge sharing among different IoT systems are still open issues to be addressed. In this paper, we propose a novel approach, the Experience-Oriented Smart Things (EOST), that utilizes deep learning and knowledge representation concept called Decisional DNA to help IoT systems acquire, represent, and store knowledge, as well as share it amid various domains where it can be required to support decisions. Decisional DNA motivation stems from the role of deoxyribonucleic acid (DNA) in storing and sharing information and knowledge. We demonstrate our approach in a set of experiments, in which the IoT systems use knowledge gained from past experience to make decisions and predictions. The presented initial results show that the EOST is a very promising approach for knowledge capture, representation, sharing, and reusing in IoT systems.

In this paper, we propose a Neural Knowledge DNA (NK-DNA)-based framework that is capable of learning from the car’s daily operations and reusing such learned knowledge in future tasks. The NK-DNA is a novel knowledge representation and reasoning approach designed to support discovering, storing, reusing, improving, and sharing knowledge among machines and computing devices. We examine our framework for drivers’ classification based on their driving behaviors. The experimental data are collected via smartphone sensors. The initial results are presented, and the direction for our future research is defined.

In this article, we introduce a novel concept combining neural network technology and Decisional DNA for knowledge representation and sharing. Instead of using traditional machine learning and knowledge discovery methods, this approach explores the way of knowledge extraction through deep learning processes based on a domain’s past decisional events captured by Decisional DNA. We compare our approach with kNN (k-nearest neighbors), logistic regression, and AdaBoost in classification tasks, and the results show that our approach is very promising with regard to the enhancement of the accuracy of knowledge-based predictions required in complex decision-making problems.

In this article, we present a novel approach utilizing Decisional DNA to help the Internet of Things capture decisional events and reuse them for decision making in future operations. The Decisional DNA is a domain-independent, standard and flexible knowledge representation structure that allows its domains to acquire, store, and share experiential knowledge and formal decision events in an explicit way. We apply this approach to our current work—SmartBike, a sensor-equipped bicycle built under the concept of Internet of Things. By using Decisional DNA and machine learning algorithms, the SmartBike is able to distinguish its user's patterns based on past riding data. The presented conceptual approach demonstrates how Decisional DNA can be applied to the Internet of Things and bring to them intelligence required by forthcoming semantic networks.

In this article, we present a concept and case study of a novel approach that generates a television (TV) user's profile utilizing principles of fuzzy logic. A user profile refers to the user's basic information, such as gender, age, and profession. The generated profile has the potential to significantly improve Digital TV (DTV), making the service smarter and more user friendly. We apply the proposed approach to our previous work that introduced decisional DNA TV, which enables TV broadcasters to suggest program choices based upon the user's past viewing habits. Decisional DNA is a domain-independent, flexible, and standard experiential knowledge repository solution that allows for knowledge to be acquired, reused, evolved, and shared in an easy and portable way. The presented conceptual approach demonstrates how fuzzy logic methods can be deployed within DNA TV through an experimental implementation that generates a user profile by capturing viewing habits.

This paper presents integration of the concept of Decisional DNA Digital TV with Smart TV. The integration provides the Digital TV viewer with smart assistance that helps to watch TV shows according to the viewer’s habit discovered through past viewing experience. Decisional DNA is a domainindependent, flexible, and standard experiential knowledge representation structure that allows its domains to acquire, reuse, evolve, and share knowledge in an easy and standard way. The presented approach demonstrates how the Decisional DNA-based systems can be integrated with Digital TV technique, and how it captures and reuses the TV viewer’s watching experience. Illustrative test of the suggested approach is presented in the paper with a set of experiments.

Knowledge and experience engineering techniques are becoming increasingly useful and popular components of hybrid integrated systems used to solve complex real-life problems in different disciplines. These techniques offer features such as learning from experience, handling noisy and incomplete data, helping with decision making, and predicting capabilities. In this article, we present a number of different applications of a multidomain knowledge representation structure called decisional DNA that can be implemented and shared for the exploitation of embedded knowledge within different technologies.

In this article, we explore an approach that integrates decisional DNA, a domain-independent, flexible, and standard knowledge repository, with digital TV in order to capture, reuse, and share viewers’ TV watching experience and preferences. Key issues in applying this approach include capturing of experience, storage and management of experience, and retrieval of experience from experience repository. We demonstrate our approach in a set of initial experiments, in which viewers’ movie watching experiences are captured and reused to support smart digital TV services.

In this article, we explore an approach that integrates Decisional DNA, a domain-independent, flexible, and standard knowledge representation structure, with robots in order to test the usability and suitability of this novel knowledge representation structure. Core issues in using this Decisional DNA–based method include capturing of knowledge, storage and indexing of knowledge, organization of the knowledge base memory, and retrieval of knowledge from memory according to current problems. We demonstrate our approach in a set of experiments in which the robots capture knowledge from their tasks and are able to reuse such knowledge in subsequent tests.