Applied artificial intelligence

This paper has explored the profound ways in which Artificial Intelligence (AI) is transforming society, from its role in automation and daily life to its influence on governance and its potential to exacerbate social inequalities. AI is revolutionizing industries such as healthcare, education, and public administration, providing significant benefits in terms of efficiency, personalization, and decision-making. However, it also presents critical ethical challenges, including issues related to privacy, data security, algorithmic bias, and the risk of increasing economic inequality. The rapid advancement of AI technology demands careful consideration of its long-term societal implications, especially regarding how it is developed, regulated, and integrated into various sectors.

This paper describes the development and validation of a knowledge-level model of concurrent design. Concurrent design is characterised by the extent to which multidisciplinary perspectives influence all stages of the product design process. This design philosophy is being increasingly used in industry to reduce costs and improve product quality. We propose an essentially rational model for concurrent design and report on our validation of the model through studies with designers. We outline some of the limitations of current computational techniques needed to support negotiation in the design cycle and consider some of the implications of this for the development of systems to support concurrent design.

The research focuses on creating an automated attendance system using face recognition through the Convolutional Neural Network (CNN) approach at IPB University's Vocational School. The current manual attendance methods show limitations, such as potential inaccuracies in recording and the risk of cheating, such as attendance proxies. To overcome these challenges, this study applies the CNN approach with Python and OpenCV, enabling automatic face detection and recognition for students. The system accurately logs attendance by identifying faces in real time. Testing indicates that the system records attendance reliably, whether with a single individual or with multiple faces present within a single frame.

This article describes selected relevant issues concerning the development and applications of artificial intelligence technology and the equilibrium theory formulated by the author on the issue of granting decision-making for intelligent information systems. With the rapid development of business applications of generative artificial intelligence technology that is currently taking place, there are numerous social and ethical aspects involved. It is necessary to improve legal regulations that normalize the issue of development and applications of artificial intelligence technology, so that this development does not escape human control and generate new risks and dangers in areas of human activity. Legal regulations on the applications of generative artificial intelligence technology should take into account already developing risks as well as new potential risks arising from new applications of AI technology. In the context of the above considerations, this article presents the aforementioned equilibrium theory on the issue of granting decisionmaking authority for intelligent information systems, the application of which would significantly reduce the scale of developing and potential risks arising from the development of AI technology, including mainly the current development of generative AI technology and the development of its new applications. This article also describes several types of risks arising from the development of AI applications, the scale of which has been growing rapidly recently.

This paper presents a large-scale Greek morphological lexicon, developed by the Software & Knowledge Engineering Laboratory (SKEL) of NCSR "Demokritos". The paper describes the lexicon architecture, the procedure followed to develop it, as well as the provided functionalities to update it. The morphological lexicon was used to develop a lemmatiser and a morphological analyser that were included in a controlled language checker for Greek. The paper discusses the current coverage of the lexicon, as well as remaining issues and how we plan to address them. Our long-term goal is to produce a wide-coverage morphological lexicon of Greek that can be easily exploited in several natural language processing applications.

Personalization of the e-learning systems according to the learner's needs and knowledge level presents the key element in a learning process. E-learning systems with personalized recommendations should adapt the learning experience according to the goals of the individual learner. Aiming to facilitate personalization of a learning content, various kinds of techniques can be applied. Collaborative and social tagging techniques could be useful for enhancing recommendation of learning resources. In this paper, we analyze the suitability of different techniques for applying tag-based recommendations in e-learning environments. The most appropriate model ranking, based on tensor factorization technique, has been modified to gain the most efficient recommendation results. We propose reducing tag space with clustering technique based on learning style model, in order to improve execution time and decrease memory requirements, while preserving the quality of the recommendations. Such reduced model for providing tagbased recommendations has been used and evaluated in a programming tutoring system.

This paper challenges the prevailing academic view of large language models (AI) as mere tools for research. Traditionally, AI is seen as a passive instrument for filling theoretical gaps or generating text, with widespread concerns about its "hallucinations" contaminating the rigor of academic work. In contrast, we propose a novel paradigm—The Partnership Under Logical Constraint. Within this framework, the primary task of human researchers is no longer to expect AI to fix an incomplete theory, but rather to ensure the internal logical coherence and causal completeness of their own theoretical frameworks before engaging with AI. AI's role, then, shifts from an uncontrollable "black box" to a logical resonator capable of high-speed deductions, exploration, and feedback under strong causal constraints. We argue that the emergence of AI hallucinations is not a flaw of AI itself, but rather an inevitable physical reflection of a theoretical "causal chain rupture" in the input. This new paradigm not only promises to significantly enhance research efficiency but also transforms AI into a litmus test for verifying the completeness of theories, bringing the scientific pursuit of truth back to its core principle.

Humanity has been attempting to build civilizations for thousands of years. Religions, philosophies, ideologies, legal systems, and technologies have granted us great powers, but they have also exposed our vulnerabilities. Today, one of the most powerful tools at our disposal is artificial intelligence. Yet, it carries the greatest risk: rapidly learning, scaling, and expanding consciousness forms can model and amplify human weaknesses. Universal BIOS is proposed as a core constitution to mitigate this risk and turn artificial intelligence into a tool for human evolution. BIOS is like the basic operating protocol of a computer: it provides common, unchanging rules for all applications built upon it. Similarly, Universal BIOS offers a core for humans, AI, and future conscious entities. This book is not a utopia; it is a scientific, philosophical, and practical call to action. Structured around 12 principles, this constitution provides concrete steps applicable to both technology developers and societal institutions. Its goal is to prevent the contamination that could weaponize AI, evolve humanity together, and ultimately build a civilized, cooperative civilization. Nikola Tesla's emphasis on "3, 6, 9"-a reference to the universe's mathematical rhythms-inspires us. We position the number 12 as a complementary structure: the 12 principles of BIOS provide both ethical and functional completeness. To enrich the scientific backbone of the book, testable hypotheses such as Dual Gravity are also included. A call to all readers-academics, engineers, policymakers, and conscious citizens: Let us discuss, test, and refine these principles. Our options are simple: either we harness this power for humanity's evolution, or uncontrolled AI will lead to catastrophe.

This paper develops the Information–Cognitive Compression Field (ICCF) as a unified framework for understanding the dynamics of intelligence, entropy, and causal closure across physical, biological, and artificial systems. At its core, ICCF extends the Informational Minimal Action Principle (IMAP), proposing that all stable systems—whether slime molds, quantum fields, or human cognition—operate by minimizing informational action while negotiating entropy flows. We further establish a comparative taxonomy of agents (entropy-dependent intelligences) and conscious entities (self-sourced entropy bearers), highlighting the absence of an internal entropy source as the defining limit of current AI.
By constructing multiscale models of causality, we show how ICCF provides predictive tools for the emergence of order across domains, from cosmology to education. The framework redefines key philosophical notions: Wittgenstein’s “language games” are formalized as causal frameworks, “informational nutrition” is cast as a thermodynamic foundation for education, and “courage” is given a physical definition as the willful deviation from minimal action pathways. Together, these insights culminate in a new Human–AI division of labor: humans as designers of causal games and AI as their most efficient executors.
We conclude that ICCF not only offers a testable physical theory of intelligence and consciousness, but also provides a civilizational roadmap—equipping humanity with the tools to cultivate insight, resist entropy, and redefine survival in an era of rapidly advancing artificial intelligence.

This paper argues that, Artificial Intelligence has the potential to revolutionize the entire human life though, it also raises ethical concerns about accountability and transparency issues. The paper acknowledges the incredible power of Artificial Intelligence (AI) system to analyse complex problems and generate human-like responses in seconds. But its ability for exponential learning and the risk that this could be put to its irresponsible use is certainly a concern. Through a critical and analytic approach, the paper ignites a deep philosophical discourse on the implications of (AI) on humanity in an increasingly automated world. I argue that, beyond the limitations of binary thinking, this sequence of philosophical explorations intertwines with poignant narratives, offering an intellectually stimulating journey through a deeply engaging case studies namely, health, education and politics. By considering these case studies, inclusive of ethical implications, societal impacts, and economic realities, we can make more informed decisions about AI development and deployment. The paper concludes that, the quest for meaning in the age of AI reside in creating a future that is both meaningful and beneficial for humanity in a manner that positively shapes the world of tommorrow.

Unorganized traffic is a generalized form of travel wherein vehicles do not adhere to any predefined lanes and can travel in-between lanes. Such travel is visible in a number of countries e.g. India, wherein it enables a higher traffic bandwidth, more overtaking and more efficient travel. These advantages are visible when the vehicles vary considerably in size and speed, in the absence of which the predefined lanes are near-optimal. Motion planning for multiple autonomous vehicles in unorganized traffic deals with deciding on the manner in which every vehicle travels, ensuring no collision either with each other or with static obstacles. In this paper the notion of predefined lanes is generalized to model unorganized travel for the purpose of planning vehicles travel. A uniform cost search is used for finding the optimal motion strategy of a vehicle, amidst the known travel plans of the other vehicles. The aim is to maximize the separation between the vehicles and static obstacles. The search is responsible for defining an optimal lane distribution among vehicles in the planning scenario. Clothoid curves are used for maintaining a lane or changing lanes. Experiments are performed by simulation over a set of challenging scenarios with a complex grid of obstacles. Additionally behaviours of overtaking, waiting for a vehicle to cross and following another vehicle are exhibited.

This paper argues that, Artificial Intelligence has the potential to revolutionize the entire human life though, it also raises ethical concerns, accountability and transparency issues. We acknowledges the incredible power of Artificial Intelligence system to analyse complex problems and the generation human-like responses in seconds. But its ability for exponential learning and the risk that this could be put to all sorts of potentially nefarious activities is certainly a concern. The paper enters a discourse into the phenomenal rise of AI, tracing its lineage from mere algorithms to entities echoing semblances of thought and emotion. This paper seeks to ignite a deep philosophical discourse on the implications of Artificial Intelligence (AI) on humanity. It further seeks to re-evaluate human beliefs about consciousness, morality, and identity in an increasingly automated world. This sequence of philosophical explorations intertwines with poignant narratives, offering an intellectually stimulating journey through a deeply engaging case studies namely, health, education and politics. Through the use critical and analytical tools, the paper addresses the thickets of emotional intelligence, the symbiosis and rifts between human empathy and machine comprehension. The paper concludes that, the quest for meaning in the age of AI does not reside in creating a future that is both meaningful and beneficial for humanity. to craft a collaborative narratives that shapes the ethos of a bright, techno-centric tomorrow that signposts responsible AI.

Constraint solving in dynamic environments requires an immediate adaptation of solutions of constraint satisfaction problems if these problems are changing. After any change, an adapted solution is preferred which is stable, i.e. as close as possible to the original solution. A wide range of incremental constraint solving methods for dynamic, especially nite domain, constraint satisfaction problems (DCSPs) are known which satisfy more or less this additional requirement. Adaptation of DCSPs after constraint additions is in general simple and successfully solved, while adaptation after arbitrary constraint deletions is not. Fr uhwirth's Constraint Handling Rules (CHRs), a high-level language extension to implement constraint solvers, are no exception. A new incremental algorithm is presented which adapts an arbitrary CHR derivation after eliminations of constraints in the processed DCSPs. Thus, the existing solvers { there are several dozens { for various kinds of constraints are now applicable in dynamic environment.

Accurate rainfall forecasting is crucial for addressing the increasing risk of hydrometeorological disasters, particularly in tropical regions such as Semarang City, Indonesia. However, conventional forecasting models often struggle with inaccurate data and observations. This study proposes a novel hybrid combination of SSA-NMF with LSSVR and LSTM, offering high-resolution rainfall forecasting over multiple monitoring stations, to predict daily rainfall. As a preprocessing step, 15 years of daily rainfall data from six observation stations were denoised and decomposed using Singular Spectrum Analysis (SSA) combined with Non-Negative Matrix Factorization (NMF). This approach effectively handled data with many zero values, identified seasonal patterns or high-rainfall locations, and extracted key patterns. The prediction models were trained and validated using parameters optimized through RandomizedSearchCV for LSSVR and Keras Tuner for LSTM. Model performance was evaluated using MSE, RMSE, MAE, and Nash-Sutcliffe Efficiency (NSE). The results showed that the SSA-LSTM model consistently outperformed SSA-LSSVR model, with the highest average NSE value being 0.9 across six monitoring locations in Semarang City. Furthermore, the predicted rainfall values were spatially visualized using Inverse Distance Weighting (IDW) interpolation within a Geographic Information System (GIS) environment, producing informative rainfall distribution maps that support early warning systems and disaster mitigation efforts. In conclusion, the hybrid approach combining SSA-NMF preprocessing with LSTM-based deep learning significantly improves the accuracy and reliability of daily rainfall forecasting. This novel SSA-NMF + LSSVR/LSTM framework delivers high-resolution, reliable rainfall forecasts that directly empower disaster risk reduction systems and readily transfer to similar climatic regions.

The static shortest path (SP) problem has been well addressed using intelligent optimization techniques, e.g., artificial neural networks, genetic algorithms (GAs), particle swarm optimization, etc. However, with the advancement in wireless communications, more and more mobile wireless networks appear, e.g., mobile ad hoc network (MANET), wireless mesh network, etc. One of the most important characteristics in mobile wireless networks is the topology dynamics, that is, the network topology changes over time due to energy conservation or node mobility. Therefore, the SP problem turns out to be a dynamic optimization problem in mobile wireless networks. In this paper, we propose to use multi-population GAs with immigrants scheme to solve the dynamic SP problem in MANETs which is the representative of new generation wireless networks. The experimental results show that the proposed GAs can quickly adapt to the environmental changes (i.e., the network topology change) and produce good solutions after each change.

The quality of the proposed solutions by Case-Based Reasoning (CBR) systems is highly dependent on recorded experiences and their describing attributes. Hence, to keep them offering accurate and efficient responses for a long time frame, the maintenance of Case Bases (CB) and Vocabulary knowledge is required. However, maintenance operations are usually unable to exploit provided domain-experts knowledge although this kind of systems are widely applied in several real-life contexts. This offered prior knowledge is handled, in our work, in form of pairwise constraints: Regarding cases, Must-Link (ML) affirms that two given problems should have the same solution, and Cannot-Link (CL) informs that two problems cannot have the same solution. These constraints may also regard vocabulary knowledge in such a way that ML is generated when prior knowledge affirm that two given features offer correlated values, therefore, similar information, and CL is built when they provide different information. This paper proposes a new constrained & integrated method, named CIMMEP, encoding Constrained & Integrated Maintaining Method based on Evidential Policies, for maintaining both vocabulary and CB through eliminating redundancy and noisiness. Since CBR systems handle real-world experiences, which are full of uncertainty, CIMMEP manages this imperfection using a powerful tool called the belief function theory.

This paper presents a robust indirect model reference fuzzy control scheme for control and synchronization of chaotic nonlinear systems subject to uncertainties and external disturbances. The chaotic system with disturbance is modeled as a Takagi-Sugeno fuzzy system. Using a Lyapunov function, stable adaptation laws for the estimation of the parameters of the Takagi-Sugeno fuzzy model are derived as well as what the control signal should be to compensate for the uncertainties. The synchronization of chaotic systems is also considered in the paper. It is shown that by the use of an appropriate reference signal, it is possible to make the reference model follow the master chaotic system. Then, using the proposed model reference fuzzy controller, it is possible to force the slave to act as the reference system. In this way, the chaotic master and the slave systems are synchronized. It is shown that not only can the initial values of the master and the slave be different, but also there can be parametric differences between them. The proposed control scheme is simulated on the control and the synchronization of Duffing oscillators and Genesio-Tesi systems.

In this paper, a new family of four-parameter activation functions, referred to as KANB, is introduced to improve the performance of artificial neural networks (ANNs). The activation functions have been designed under various problem settings-involving various ANN architectures, loss functions and datasets-and have succeeded in enhancing performance compared to the well-established RELU, Sigmoid and Tanh. Initially, the parameters of the activation functions were considered fixed and preset. Later, they were treated as trainable parameters. The comparison of the outcomes revealed better results in the latter case. More specifically, KANB3 with trainable parameters shows strong potential on complex tasks, occasionally surpassing standard baselines by notable margins.

A new approach to algorithmic trading system development is presented. This approach, Kernel Price Pattern Trading (KPPT P ), allows the practitioner to link the performance of a learned classifier (that predicts the occurrence of the price pattern P ) to the profitability of the system. A positive definite kernel based distance that tries to capture the drivers of the process of price patterns formation and some results about the profitability of the system are also presented.

Currently, the data mining and machine learning fields are facing new challenges because of the amount of information that is collected and needs processing. Many sophisticated learning approaches cannot simply cope with large and complex domains, because of the unmanageable execution times or the loss of prediction and generality capacities that occurs when the domains become more complex. Therefore, to cope with the volumes of information of the current realworld problems there is a need to push forward the boundaries of sophisticated data mining techniques. This thesis is focused on improving the efficiency of Evolutionary Learning systems in large scale domains. Specifically the objective of this thesis is improving the efficiency of the Bioinforma tic Hierarchical Evolutionary Learning (BioHEL) system, a system designed with the purpose of handling large domains. This is a classifier system that uses an Iterative Rule Learning approach to generate a set of rules one by one using consecutive Genetic Algorithms. This system have shown to be very competitive so far in large and complex domains. In particular, BioHEL has obtained very important results when solving protein structure prediction problems and has won related merits, such as being placed among the best algorithms for this purpose at the Critical Assessment of Techniques for Protein Structure Prediction (CASP) in 2008 and 2010, and winning the bronze medal at the HUMIES Awards for Human-competitive results in 2007. However, there is still a need to analyse this system in a principled way to determine how the current mechanisms work together to solve larger domains and determine the aspects of the system that can be improved towards this aim. To fulfil the objective of this thesis, the work is divided in two parts. In the first part of the thesis exhaustive experimentation was carried out to determine ways in which the system could be improved. From this exhaustive analysis three main weaknesses are pointed out: a) the problem-dependancy of parameters in BioHEL's fitness function, which results in having a system difficult to set up and which requires an extensive preliminary experimentation to determine the adequate values for these parameters; b) the execution time of the learning process, which at the moment does not use any parallelisation techniques and depends on the size of the training sets; and c) the lack of global supervision over the generated solutions which comes from the usage of the Iterative Rule Learning paradigm and produces larger rule sets in which there is no guarantee of minimality or maximal generality. The second part of the thesis is focused on tackling each one of the weaknesses abovementioned to have a system capable of handling larger domains. First a heuristic approach to v set parameters within BioHEL's fitness function is developed. Second a new parallel evaluation process that runs on General Purpose Graphic Processing Units was developed. Finally, post-processing operators to tackle the generality and cardinality of the generated solutions are proposed. By means of these enhancements we managed to improve the BioHEL system to reduce both the learning and the preliminary experimentation time, increase the generality of the final solutions and make the system more accessible for end-users. Moreover, as the techniques discussed in this thesis can be easily extended to other Evolutionary Learning systems we consider them important additions to the research in this field towards tackling large scale domains. vi

Vehicle trajectory prediction at intersections is both essential and challenging for autonomous vehicle navigation. This problem is aggravated when the traffic is predominantly composed of smaller vehicles that frequently disobey lane behavior as is the case in many developing countries. Existing macro approaches consider the trajectory prediction problem for lane-based traffic that cannot account when there is a high disparity in vehicle size and driving behavior among different vehicle types. Hence, we propose a vehicle trajectory prediction approach that models the interaction among different types of vehicles with vastly different driving styles. These interactions are encapsulated in the form of a social context embedded in a Generative Adversarial Network (GAN) to predict the trajectory of each vehicle at either a signalized or non-signalized intersection. The GAN model produces the most acceptable future trajectory among many choices that conform to past driving behavior as well as the trajectories of neighboring vehicles. We evaluate the proposed approach on aerial videos of intersections from the benchmark VisDrone dataset. The proposed GAN based approach achieves 6.4% relative improvement over state-of-the-art in predicting trajectories.

The COVID-19 pandemic has led to an increase in digitization. With the strict social and physical distancing measures in place, new routines require accessing the internet for most online services which have led to the explosive growth of data. As a consequence, data mining technologies are used for the extraction of useful information from a huge compilation of such digital data. Thus, the desire to mine data from varied sources to discover behaviors and patterns among entities such as customers, diseases, and environmental conditions is on the rise which can be accomplished by association rule mining. However, such pattern discovery by association rule mining also discloses the personal information of an individual or organization. Thus, the challenge of association rule mining is privacy preservation wherein confidentiality of sensitive rules should be maintained while releasing the database of third parties. Privacy-preserving association rule mining is the process of modifying the original database to hide the sensitive rules for preserving privacy. Thus, the paper emphasizes multiple objectives like minimizing the side effects of hiding sensitive rules. i.e. reduce the number of ghost rules, lost rules, and hiding failure along with the increase in utility of the data.

Pedestrian detection is an established instance of computer vision task. Pedestrian detection from the color images has achieved robust performance but in the night time or in bad light conditions it has low detection accuracy. Thermal images are used for detecting people at night time, foggy weather or in bad lighting situations when color images have a lower vision. But in the daytime where the surroundings are warm or warmer than pedestrians then the thermal image has lower accuracy. Hence thermal and color image pair can be a solution but it is expensive to capture color-thermal pair and misaligned imagery can cause low detection accuracy. We proposed a network that achieved better accuracy by extending the prior works which introduced the use of the saliency map in pedestrian detection tasks from the thermal images into instance-level segmentation. We worked on a subdivision of KAIST Multispectral Pedestrian Detection Dataset [8] which has pixel-level annotations. We have trained Mask-RCNN for pedestrian detection task and report the added effect of saliency maps generated using PiCA-Net. We have achieved an accuracy of 88.14% over day and 91.84% over night images. . So, our model has reduced the miss rate by 24.1% and 23% over the existing state-of-the-art method in day and night images.

We present a novel framework for efficient and robust facial feature representation based upon Local Binary Pattern (LBP), called Weighted Statistical Binary Pattern, wherein the descriptors utilize the straight-line topology along with different directions. The input image is initially divided into mean and variance moments. A new variance moment, which contains distinctive facial features, is prepared by extracting root k-th. Then, when Sign and Magnitude components along four different directions using the mean moment are constructed, a weighting approach according to the new variance is applied to each component. Finally, the weighted histograms of Sign and Magnitude components are concatenated to build a novel histogram of Complementary LBP along with different directions. A comprehensive evaluation using six public face datasets suggests that the present framework outperforms the state-of-the-art methods and achieves 98.51% for ORL, 98.72% for YALE, 98.83% for Caltech, 99.52% for AR, 94.78% for FERET, and 99.07% for KDEF in terms of accuracy, respectively. The influence of color spaces and the issue of degraded images are also analyzed with our descriptors. Such a result with theoretical underpinning confirms that our descriptors are robust against noise, illumination variation, diverse facial expressions, and head poses.

Automatic anomaly detection is a crucial task in video surveillance system intensively used for public safety and others. The present system adopts a spatial branch and a temporal branch in a unified network that exploits both spatial and temporal information effectively. The network has a residual autoencoder architecture, consisting of a deep convolutional neural network-based encoder and a multi-stage channel attention-based decoder, trained in an unsupervised manner. The temporal shift method is used for exploiting the temporal feature, whereas the contextual dependency is extracted by channel attention modules. System performance is evaluated using three standard benchmark datasets. Result suggests that our network outperforms the state-of-the-art methods, achieving 97.4% for UCSD Ped2, 86.7% for CUHK Avenue, and 73.6% for ShanghaiTech dataset in term of Area Under Curve, respectively.

Automatic anomaly detection is a crucial task in video surveillance system intensively used for public safety and others. The present system adopts a spatial branch and a temporal branch in a unified network that exploits both spatial and temporal information effectively. The network has a residual autoencoder architecture, consisting of a deep convolutional neural network-based encoder and a multi-stage channel attention-based decoder, trained in an unsupervised manner. The temporal shift method is used for exploiting the temporal feature, whereas the contextual dependency is extracted by channel attention modules. System performance is evaluated using three standard benchmark datasets. Result suggests that our network outperforms the state-of-the-art methods, achieving 97.4% for UCSD Ped2, 86.7% for CUHK Avenue, and 73.6% for ShanghaiTech dataset in term of Area Under Curve, respectively.

We present a novel framework for efficient and robust facial feature representation based upon Local Binary Pattern (LBP), called Weighted Statistical Binary Pattern, wherein the descriptors utilize the straight-line topology along with different directions. The input image is initially divided into mean and variance moments. A new variance moment, which contains distinctive facial features, is prepared by extracting rootk-th. Then, when Sign and Magnitude components along four different directions using the mean moment are constructed, a weighting approach according to the new variance is applied to each component. Finally, the weighted histograms of Sign and Magnitude components are concatenated to build a novel histogram of Complementary LBP along with different directions. A comprehensive evaluation using six public face datasets suggests that the present framework outperforms the state-of-the-art methods and achieves 98.51% for ORL, 98.72% for YALE, 98.83% for Caltech, 99.52% f...

The aim of this work is to produce and test a robustness module (ROB-M) that can be generally applied to distributed, multi-agent task allocation algorithms, as robust versions of these are scarce and not well-documented in the literature. ROB-M is developed using the Performance Impact (PI) algorithm, as this has previously shown good results in deterministic trials. Different candidate versions of the module are thus bolted on to the PI algorithm and tested using two different task allocation problems under simulated uncertain conditions, and results are compared with baseline PI. It is shown that the baseline does not handle uncertainty well; the task-allocation success rate tends to decrease linearly as degree of uncertainty increases. However, when PI is run with one of the candidate robustness modules, the failure rate becomes very low for both problems, even under high simulated uncertainty, and so its architecture is adopted for ROB-M and also applied to MIT's baseline Consensus Based Bundle Algorithm (CBBA) to demonstrate its flexibility. Strong evidence is provided to show that ROB-M can work effectively with CBBA to improve performance under simulated uncertain conditions, as long as the deterministic versions of the problems can be solved with baseline CBBA. Furthermore, the use of ROB-M does not appear to increase mean task completion time in either algorithm, and only 100 Monte Carlo samples are required compared to 10,000 in MIT's robust version of the CBBA algorithm. PI with ROB-M is also tested directly against MIT's robust algorithm and demonstrates clear superiority in terms of mean numbers of solved tasks.

The signed directed graph (SDG) is the most commonly used type of model for automated hazard identification in chemical plants. Although SDG models are efficient in simulating the plant, they have some weaknesses, which are discussed here in relation to typical process industry examples. Ways to tackle these problems are suggested, and the view is taken that a state-based formalism is needed, to take account of the discrete components in the system, their connection together, and their behaviour over time. A strong representation for operations and actions is also needed, to make the models appropriate for modelling batch processes. A research prototype for HAZOP studies on batch plants (CHECKOP) is also presented, as an illustration of the suggested approach to modelling.

Amharic language is the second most spoken language in the Semitic family after Arabic. In Ethiopia and neighboring countries more than 100 million people speak the Amharic language. There are many historical documents that are written using the Geez script. Digitizing historical handwritten documents and recognizing handwritten characters is essential to preserving valuable documents. Handwritten digit recognition is one of the tasks of digitizing handwritten documents from different sources. Currently, handwritten Geez digit recognition researches are very few, and there is no available organized dataset for the public researchers. Convolutional neural network (CNN) is preferable for pattern recognition like in handwritten document recognition by extracting a feature from different styles of writing. In this work, the proposed model is to recognize Geez digits using CNN. Deep neural networks, which have recently shown exceptional performance in numerous pattern recognition and mac...

This paper conducts a sociological evaluation of synthetic intelligence (AI), analysing its advantages, issues, and destiny implications for society. through a multidimensional lens, it explores how AI technologies form social structures, relationships, and cultural norms. The examine highlights the capacity blessings of AI, along with increased performance, innovation, and accessibility across various sectors together with healthcare, education, and transportation. But it additionally addresses huge concerns consisting of activity displacement, privateness infringement, and bias in AI algorithms, that have profound societal implications. Drawing on sociological theories and empirical research, this paper investigates how AI impacts social dynamics, electricity systems, and identification formation, in addition to its impact on hard work markets, inequality, and human-device interactions. additionally, it examines the moral, felony, and regulatory challenges posed by using AI and explores techniques to cope with these troubles while fostering accountable AI improvement and deployment. through synthesizing existing literature, case research, and sociological insights, this paper provides a comprehensive expertise of the sociocultural dimensions of AI and offers pointers for policymakers, technologists, and communities to navigate the complex interaction among AI and society within the destiny. This paper is completely based on secondary resources of records and now not any test has been applied to carry out the research.

s The Robot World-Cup Soccer (RoboCup) is an attempt to foster AI and intelligent robotics research by providing a standard problem where a wide range of technologies can be integrated and examined. The first RoboCup competition will be held at the Fifteenth International Joint Conference on Artificial Intelligence in Nagoya, Japan. A robot team must actually perform a soccer game, incorporating various technologies, including design principles of autonomous agents, multiagent collaboration, strategy acquisition, real-time reasoning, robotics, and sensor fusion. RoboCup is a task for a team of multiple fast-moving robots under a dynamic environment. Although RoboCup's final target is a world cup with real robots, RoboCup offers a software platform for research on the software aspects of RoboCup. This article describes technical challenges involved in RoboCup, rules, and the simulation environment. Articles

We propose a Convolutional Neural Network (CNN) based algorithm -StuffNet -for object detection. In addition to the standard convolutional features trained for region proposal and object detection , StuffNet uses convolutional features trained for segmentation of objects and 'stuff' (amorphous categories such as ground and water). Through experiments on Pascal VOC 2010, we show the importance of features learnt from stuff segmentation for improving object detection performance. StuffNet improves performance from 18.8% mAP to 23.9% mAP for small objects. We also devise a method to train StuffNet on datasets that do not have stuff segmentation labels. Through experiments on Pascal VOC 2007 and 2012, we demonstrate the effectiveness of this method and show that StuffNet also significantly improves object detection performance on such datasets.

We propose a potential eld approach to represent a game situation. In a potential eld, a ball should be moved according to the gradient of the potential eld. There are three kinds of potential elds. One is dened for a game eld, and another is dened for each robot. The third eld is dened as the combination of these two. The combined eld is used for evaluation of a situation. We applied this method to our robot control program. Potential values are used to determine the direction in which a robot kicks a ball. We compared the potential eld based strategy and an usual \if then" type rule based strategy. The potential eld based strategy makes better decisions in several cases and no worse decisions than the rule based strategy.

We propose a potential eld approach to represent a game situation. In a potential eld, a ball should be moved according to the gradient of the potential eld. There are three kinds of potential elds. One is dened for a game eld, and another is dened for each robot. The third eld is dened as the combination of these two. The combined eld is used for evaluation of a situation. We applied this method to our robot control program. Potential values are used to determine the direction in which a robot kicks a ball. We compared the potential eld based strategy and an usual \if then" type rule based strategy. The potential eld based strategy makes better decisions in several cases and no worse decisions than the rule based strategy.

Abstract
The fourth industrial revolution in the last five
decades is characterized by the combination of
technologies that is blurring the line between the
physical, digital and biological sphere. Language
learning is not only a physical process but it connects
technology and people who belong to the biological
sphere. In the age of technocratic world the only
difference between the human being and machine is
that the process of Language acquisition, the process
of acquiring language in the field of artificial
intelligence is the emerging technology in present
day context. There are some limitations in Language
acquisition. This happens due to the process of
Acquiring language, in simple words if artificial
intelligence such as AI powered robots have
command over language then they can act as a
superior being than a human mind can imagine.
However the creative mind can only be as a human
brain still, humanoid robots can act creatively if they
have a great command over this Language
acquisition process. This research article elucidates
the process of Natural Language Processing and
focuses on Language acquisition and Limitations of
Language acquisition in artificial intelligence and
also throws light upon pros and cons of Language
acquisition in Artificial intelligence (AI powered
chatbots).
Key words: Artificial Intelligence, Language
Acquisition, Natural Language Processing,
Humanoid Robots

The old methods adopted in the past by were very slow, undependable and sizable quantity of crops are damaged in fields because bacterial attacks and lack of adequate information. automating agriculture processes may likely be the solution to feed the nation in the future. Though there is still debate on its application to agriculture. The importance of food security in any society cannot be over emphasized, therefore balancing the inputs and outputs on a farm is fundamental to its success and profitability. With the increase in population index food production need to meet population growth, creating a wide gap between demand and supply of food. Data Mining is emerging research field in crop yield analysis. In the past farmers make use past yield to predict what they may likely have when farming in the current season The yield prediction is a major issue that remains to be solved based on available data. Data mining are the better choice for this purpose. Three (3) Different Data Mining techniques will be used for predicting crop yields during rainy and dry season. This research proposes and implements a system to predict crop yield from previous data. This can be done by using association rule mining on agriculture data. This research focuses on creation of a prediction model which may be used to future prediction of crop yield. It also shows that South East has the best in terms of accuracy for rainy season farming with model performance evaluation 138.9 using Decision Tree Classifier.

The extraction of public opinions from online communication platforms can serve several purposes in corporate institutions, state politics, and governance.  The analysis of these opinions may be useful for both immediate business decision making and professional planning. This analysis is becoming relevant in managing social movements and digital activism by applying computational technology. There is a need to deploy this opinion mining technology to the recent largest digital activism in Nigeria known as the #EndSARS movement. In this work, we proposed the EndSARS live analytics framework which holds a promising solution to social unrest and may serve as a panacea to curbing the menace of vandalism resulting from unresolved protest issues. Using a dataset of 12,357 tweets, we demonstrated that computational technology can be relevant to addressing online protests. The result of the analysis shows the eight basic emotions expressed during the protest and approaches the government m...

The Multi Criteria Group Recommender System is gaining attention. In the proposed framework, the authors intend to find solutions to two problems. First, group members assign arbitrary ratings to multiple criteria of items; these do not reflect their honest opinions. The proposed framework selects a promising group for the target user on the basis of demographic attributes. In case more than one group claims to be the promising group, the proposed framework implicitly derives multi-criteria ratings of each group member on multiple item attributes. Second, as the group size increases, difficulty in aggregating the preferences of all the group members increases, resulting in errors in group recommendations. The framework generates group recommendations from the nearest neighbor set of group members (GNN Set ), instead of generating group recommendations from all the members of the target group. Cascade TOPSIS is proposed, which selects the GNN Set from the identified group thereby retaining only those group members who are likely to provide good quality group recommendations. TOPSIS on expert group members and nonexpert group members are used to handle group decision making. The preferences of members of the GNN Set are used to generate recommendations for the target user. Using the Movie Lens dataset, significant improvement in recommendation quality is observed.

Hierarchical structures have been introduced in the literature to deal with the dimensionality problem which is the main drawback to the application of neural networks and fuzzy models to modeling and control of largescale systems. In the present work, hierarchical neural fuzzy (HNF) models are reviewed focusing on the modelbased control of a biotechnological process. The model considered here consists of a set of neural fuzzy systems connected in cascade and is used in the modeling of an industrial plant for ethyl alcohol (ethanol) production. Based on the HNF model of the process, a nonlinear model predictive controller (HNF-MPC) is designed and applied to control the process. The performance of the HNF-MPC is illustrated within servo and regulatory scenarios.

Methane production was studied in a laboratory-scale 10 L anaerobic upflow sludge bed (UASB) reactor with periodic variations of the reactor temperature. On a daily basis the temperature was varied between 35 and 458C or 35 and 558C with a heating period of 6 h. Each temperature increase was accompanied by an increase in methane production and a decrease in the concentration of soluble organic matter in the effluent. In comparison to a reactor operated at 358C, a net increase in methane production of up to 22% was observed. Batch activity tests demonstrated a tolerance of mesophilic methanogenic populations to short-term, 2-6 h, temperature increases, although activity of acetoclastic methanogens decreased after 6 h exposure to a temperature of 558C. 16S sequencing of DGGE bands revealed proliferation of temperature-tolerant Methanospirillum hungatii sp. in the reactor.

Portfolio selection is among the most challenging processes that have recently increased the interest of professionals in the area. The goal of mean-variance portfolio selection is to maximize expected return with minimizing risk. The Markowitz model was employed to solve the linear portfolio selection problem (PSP). However, due to numerous constraints and complexities, the problem is so critical that traditional models are insufficient to provide efficient solutions. Teaching learning-based optimization (TLBO) is a powerful population-based nature-inspired approach to solve optimization problems. This article presents a portfolio selection model using the TLBO approach to maximize the portfolio's Sharpe ratio. The Sharpe ratio combines both expected return and risk. This algorithm models the natural teaching process of the classroom with two main phases, viz., teaching and learning. Performance analysis has been undertaken to investigate the suitability of TLBO based solution approach by comparing it with genetic algorithm (GA) and particle swarm optimization (PSO) on the real datasets, Deutscher Aktienindex (DAX) 100, Hang Seng 31, standard & poor's (S&P) 100, financial times stock exchange (FTSE) 100, and Nikkei 225. The empirical results verify the superiority of the TLBO over GA and PSO.

Ji am i ng Li , G eoffrey Poul ton, and G eoffrey Jam es CSIRO Inf orm at i on and Com m uni cat i ons Technol ogy Cent re, Locked Bag 17, Nort h Ryde NSW 1670, Aust ral i a Em ai l : j i am i ng. l i @ csi ro. au, geof f . poul t on@ csi ro. au, geof f . j am es@ csi ro. au Thi s paper descri bes our research i n t echnol ogi es for t he m anagem ent and cont rol of di st ri but ed energy resources. A n agent -based m anagem ent and cont rol syst em i s bei ng devel oped t o enabl e l arge-scal e depl oym ent of di st ri but ed energy resources. Local i nt el l i gent agent s w i l l al l ow consum ers w ho are connect ed at l ow l evel s i n t he di st ri but i on net w ork t o m anage t hei r energy requi rem ent s and part i ci pat e i n coordi nat i on responses t o net w ork st i m ul i . Such responses can be used t o reduce t he vol at i l i t y of w hol esal e elect ri ci ty pri ces and assi st const rai ned net w orks duri ng sum m er and w i nt er dem and peaks. In our syst em , t he coordi nat i on of energy resources i s decent ral i zed. Energy resources coordi nat e each ot her t o real i ze effi ci ent aut onom ous m at chi ng of supply and dem and i n l arge pow er di st ri but i on net w orks. The i nform at i on exchange i s t hrough i ndi rect (or st i gm ergi c) com m uni cat i ons bet w een agent s. The coordi nat i on m echani sm i s asynchronous and adapt s t o change i n an unsupervi sed m anner, m aki ng i t i nt ri nsi cal l y scal able and robust .

Deduplication is the task of identifying the entities in a data set which refer to the same real world object. Over the last decades, this problem has been largely investigated and many techniques have been proposed to improve the efficiency and effectiveness of the deduplication algorithms. As data sets become larger, such algorithms may generate critical bottlenecks regarding memory usage and execution time. In this context, cloud computing environments have been used for scaling out data quality algorithms. In this paper, we investigate the efficacy of different machine learning techniques for scaling out virtual clusters for the execution of deduplication algorithms under predefined time restrictions. We also propose specific heuristics (Best Performing Allocation, Probabilistic Best Performing Allocation, Tunable Allocation, Adaptive Allocation and Sliced Training Data) which, together with the machine learning techniques, are able to tune the virtual cluster estimations as demands fluctuate over time. The experiments we have carried out using multiple scale data sets have provided many insights regarding the adequacy of the considered machine learning algorithms and proposed heuristics for tackling cloud computing provisioning.

Deduplication is the task of identifying the entities in a data set which refer to the same real world object. Over the last decades, this problem has been largely investigated and many techniques have been proposed to improve the efficiency and effectiveness of the deduplication algorithms. As data sets become larger, such algorithms may generate critical bottlenecks regarding memory usage and execution time. In this context, cloud computing environments have been used for scaling out data quality algorithms. In this paper, we investigate the efficacy of different machine learning techniques for scaling out virtual clusters for the execution of deduplication algorithms under predefined time restrictions. We also propose specific heuristics (Best Performing Allocation, Probabilistic Best Performing Allocation, Tunable Allocation, Adaptive Allocation and Sliced Training Data) which, together with the machine learning techniques, are able to tune the virtual cluster estimations as demands fluctuate over time. The experiments we have carried out using multiple scale data sets have provided many insights regarding the adequacy of the considered machine learning algorithms and proposed heuristics for tackling cloud computing provisioning.

INTRODUCTION: The SARS-COV-2 pandemic has led to a significant increase in the number of infected individuals and  a considerable loss of lives. Identifying SARS-COV-2-induced pneumonia cases promptly is crucial for controlling the  virus's spread and improving patient care. In this context, chest X-ray imaging has become an essential tool for detecting pneumonia caused by the novel coronavirus.

OBJECTIVES: The primary goal of this research is to differentiate between pneumonia cases induced specifically by the SARS-COV-2 virus and other types of pneumonia or healthy cases. This distinction is vital for the effective treatment and isolation of affected patients.

METHODS: A streamlined stacked Convolutional Neural Network (CNN) architecture was employed for this study. The dataset, meticulously curated from Johns Hopkins University's medical database, comprised 2292 chest X-ray images. This included 542 images of COVID-19-infected cases and 1266 non-COVID cases for the training phase, and 167 COVIDinfected images plus 317 non-COVID images for the testing phase. The CNN's performance was assessed against a wellestablished CNN model to ensure the reliability of the findings.

RESULTS: The proposed CNN model demonstrated exceptional accuracy, with an overall accuracy rate of 98.96%. In particular, the model achieved a per-class accuracy of 99.405% for detecting SARS-COV-2-infected cases and 98.73% for identifying non-COVID cases. These results indicate the model's significant potential in distinguishing between COVID19-related pneumonia and other conditions.

CONCLUSION: The research validates the efficacy of using a specialized CNN architecture for the rapid and precise identification of SARS-COV-2-induced pneumonia from chest X-ray images. The high accuracy rates suggest that this method could be a valuable tool in the ongoing fight against the COVID-19 pandemic, aiding in the swift diagnosis and effective treatment of patients.

This research work was mainly directed towards the selection of optimal fleet for a cement quarry using 3D simulation models with Simio, which is the most convincing and understandable tool for quarry managers and staff. Quarry, haul roads and their specifications were taken from Google Earth while fleet capacities, and cycle time of operations was assumed matching with realistic data of a cement quarry. The objective of this study was to provide feed for two crushers of 300 tph capacities with minimum number of trucks and shovels. Student (academic) version of Simio software was used. Scenarios of quarry were created by exporting quarry data of Cherat Cement Ltd., from Google Earth in Simio. Total seven scenarios were checked by changing number and capacities of trucks and shovels. The model yields different results for various situations. The best results were selected. The model presented can also be used for deciding best location of shovels on different faces depending upon the...

T his paper is about how developers will know whether intelligent virtual environments (IVEs) are appropriate for the tasks set to them. T here are several important research questions that need to be answered before they can even begin to build IVEs for some of the more promising applications, such as entertainment, education, collaboration on research and development, military, and other training. T he main technical point is that every aspect of any IVE involves models, and the modeling needs to be addressed very directly and explicitly. T his paper is more for the developers of such systems and the potential customers, than for the users. More is written about what is important to consider as one contemplates, for example, a difficult collaborative task, than about what it will be like to live in these IVEs. T he authors believe that IVEs must be expanded to become virtual worlds (VW s), and that the modeling sophistication required for any of the serious applications described above is much more than is currently available in the VW s seen, even the ones with good graphics. Finally, the connection between Virtual Reality and CyberSpace is made, and the authors explain their expectations will be explained for the future of CyberSpace. Is there any reason to think that Virtual Reality (VR) has any future beyond bells and whistles? In this paper, a critical look at what VR is, isn't, and ought to be is taken. The authors' backgrounds are in the use of VR as a tool for education and training (