Less is more approach in optimization: a road to artificial intelligence

Cancer centers have an urgent and unmet clinical and research need for AI that can guide patient management. A core component of advancing cancer treatment research is assessing response to therapy. Doing so by hand, for example, as per RECIST or RANO criteria, is tedious and time-consuming, and can miss important tumor response information. Most notably, the prevalent response criteria often exclude lesions, the non-target lesions, altogether. We wish to assess change in a holistic fashion that includes all lesions, obtaining simple, informative, and automated assessments of tumor progression or regression. Because genetic sub-types of cancer can be fairly specific and patient enrollment in therapy trials is often limited in number and accrual rate, we wish to make response assessments with small training sets. Deep neuroevolution (DNE) is a novel radiology artificial intelligence (AI) optimization approach that performs well on small training sets. Here, we use a DNE parameter search to optimize a convolutional neural network (CNN) that predicts progression versus regression of metastatic brain disease. We analyzed 50 pairs of MRI contrast-enhanced images as our training set. Half of these pairs, separated in time, qualified as disease progression, while the other 25 image pairs constituted regression. We trained the parameters of a CNN via "mutations" that consisted of random CNN weight adjustments and evaluated mutation "fitness" as summed training set accuracy. We then incorporated the best mutations into the next generation's CNN, repeating this process for approximately 50,000 generations. We applied the CNNs to our training set, as well as a separate testing set with the same class balance of 25 progression and 25 regression cases. DNE achieved monotonic convergence to 100% training set accuracy. DNE also converged monotonically to 100% testing set accuracy. We have thus shown thatDNE can accurately classify brain metastatic disease progression versus regression. Future work will extend the input from 2D image slices to full 3D volumes, and include the category of "no change." We believe that an approach such as ours can ultimately provide a useful and informative complement to RANO/RECIST assessment and volumetric AI analysis.

Software defined network (SDN) is a promising technology which can reduce network management complexity through the decoupling of the control plane and data plane. Due to large number of switches in the data plane, distributed and multiple controllers are necessary in the control plane for managing the switches. The switch controller mapping strategy for identifying the mapping relationships between the switch and controller is crucial in order to optimize the network performance. Considering the dynamics of the network behavior, it is quite important and challenging to develop models to reflect the network topology dynamics and to propose method for solving the long-term network performance optimization. Inspired by the recent advances in Artificial Intelligence (AI), in this paper, we propose a Deep Reinforcement Learning (DRL) based strategy for solving the switch controller mapping problem. A DRL based mapping strategy is proposed, in which Markov Decision Process (MDP) formulation is devised and Deep $Q$ -network (DQN) is proposed to achieve the maximization of long-term system performance by leveraging network latency, load balancing and system stability. Extensive simulations show that the DQN based algorithm can achieve the best system stability results while maintaining moderate switch controller latency and system equilibrium performance comparing with the optimization which only considers current system performance for switch controller mapping decision, and the optimization approaches which generate mapping decisions purely based on latency or load balancing separately.

With the development of research in the Supervised learning field, artificial intelligence (AI) is extremely popular within in the health service. Only a few Deep Learning techniques, nevertheless, are able to find solutions within real-time applications. Data is the main worry in these types of valid medical systems, and the dentistry dataset had the same issue. The most accurate and optimal approach is presented in this study. 5 distinct datasets have been identified and assigned classifications. We propose a faster human brain using a fully convolutional model in this study. Convolution layer is a useful method in reinforcement learning for overcoming obstacles like picture segment and appropriate categorization with high levels of detection, as well as for accelerating convergence. lessens the local computation required for each outer loop. The Dense net model exhibits excellent classification in original picture classification issues when there is a large dataset, as we discovered from the current system. We have discovered three distinct maladies linked to various types of holes, including frontal gums cavities, inner pore gums, and esophageal cancer, which will distinguish between normal and diseased incisors in the sensing of teeth maladies and use this network design framework on real medical studies at various levels. To incorporate it into the web application, the created ML tree is being stored and translated into a JSON file format. The online application will serve as just an interface for identifying teeth with potential problems.

Load frequency control (LFC) has recently gained importance due to the increasing integration of wind energy in contemporary power systems. Hence, several power system models, control techniques, and controllers have been developed to improve the efficiency, resilience, flexibility, and economic feasibility of LFC. Critical factors, such as energy systems, resources, optimization approaches, resilience, and transient stability have been studied to demonstrate the uniqueness of the proposed design. This paper examines the most recent advances in LFC techniques for wind-based power systems. Moreover, the use of classical, artificial intelligence, model predictive control, sliding mode control, cascade controllers, and other newly designed and adopted controllers in the LFC area is thoroughly examined. Statistical analysis and a comparison table are used to evaluate the advantages, disadvantages, and applications of various controllers. Finally, this paper presents a comprehensive overview of contemporary and other widely used soft computing tools for the LFC issue. This detailed literature review will assist researchers in overcoming the gap between current progress, application, limitations, and future developments of wind energy in LFC.

The papers in this special section focus on advanced signal processing and artificial intelligence (AI) technologies for industrial Big Data (IBD) powered by Industry 4.0. Modern industry has evolved from the traditional manufacturing industry to digital and intelligent industry. Huge amount of complex real-time data are generated from the thousands of industrial sensors in physical and man-made environments. Industrial big data (IBD) afford us an unprecedented opportunity to obtain an in-depth understanding of Internet of Things and facilitate data-driven approaches for industrial optimization and scheduling. The papers in this section collect the latest ideas and research on advanced signal processing and artificial intelligence (AI) technologies for IBD.

Connecting aluminum alloys AA6061 and AA7075 presents significant challenges due to differences in thermal behaviors and metallurgical characteristics, often causing issues like cracking or warping during welding. Gas Metal Arc Welding (GMAW) is commonly used for aluminum alloys, but optimizing welding parameters for high-quality joints remains complex. Traditional methods are often inefficient and inadequate in multi-objective scenarios. Recent advancements in artificial intelligence (AI) offer promising alternatives, but applying AI in GMAW optimization for dissimilar aluminum alloys is still in the early stages. This study uses a hybrid Artificial Neural Network-Genetic Algorithm (ANN-GA) model to optimize GMAW parameters for these alloys, addressing gaps in traditional approaches. The ANN model, trained on experimental data, predicts tensile strength and hardness, while GA optimizes key welding parameters, including current, speed, and wire feed rate, to improve joint performance. The ANN-GA model achieved optimal settings, reaching a peak tensile strength of 237.47 MPa and maximum hardness of 98.72 HV, substantially enhancing the mechanical properties of the welded joints. These findings underscore the effectiveness of the hybrid ANN-GA approach in GMAW process optimization for dissimilar aluminum alloys. This study advances welding technology and establishes a strong framework for applying AI-driven optimization techniques to complex manufacturing processes. These insights provide valuable guidance for improving weld quality in industrial applications and pave the way for further integration of AI techniques to enhance welding performance

Various forms and symptoms of corona virus have been observed in human body especially in heart, chest and affects the respiratory system. In the initial phase, RT-PCR examination is applied to monitor the target disease, but suffers from low sensitivity and a laborious process. Apart from this, another mechanism for corona virus detection involves the analysis the CT image has become an imperative device for clinical judgment. However, manual investigation of such disease in numerous amounts of images is not the optimal approach. Additionally, recent advancement in artificial intelligence techniques have assisted medical diagnosis to identify the virus in a standard environment. In this work, the potential of such intelligence methods is analyzed and extended by considering the optimal feature extraction capability and proposes a hybrid approach in which three universal architectures namely: Inception V4, DenseNet 201 and Xception have been utilized which not only classify the corona virus disease but may also provide a pathway to apply similar method in other medical diagnosis.

In an effort to improve customization for today's highly competitive global marketplace, many companies are utilizing product families and platform-based product development to increase variety, shorten lead times, and reduce costs. The key to a successful product family is the product platform from which it is derived either by adding, removing, or substituting one or more modules to the platform or by scaling the platform in one or more dimensions to target specific market niches. This nascent field of engineering design has matured rapidly in the past decade, and this paper provides a comprehensive review of the flurry of research activity that has occurred during that time to facilitate product family design and platform-based product development for mass customization. Techniques for identifying platform leveraging strategies within a product family are reviewed along with metrics for assessing the effectiveness of product platforms and product families. Special emphasis is placed on optimization approaches and artificial intelligence techniques to assist in the process of product family design and platform-based product development. Web-based systems for product platform customization are also discussed. Examples from both industry and academia are presented throughout the paper to highlight the benefits of product families and product platforms. The paper concludes with a discussion of potential areas of research to help bridge the gap between planning and managing families of products and designing and manufacturing them.

Bentham Science - STM publisher of online and print journals, and related print/online book series. Bentham Science answers the information needs of scientists in the fields of pharmaceutical, biomedical, medical, engineering, technology, computer and social sciences.

Downtime related to excessive vessel motion and/or mooring line forces caused by environmental conditions is an important parameter for designing new terminals and ports in addition to planning offshore operations such as dredging, structure installations, etc. The present paper addresses an advanced approach using artificial intelligence in order to estimate the downtime in a more realistic manner. This approach is compared with conventional methods used in the industry and is applied for a number of terminal projects presented in this paper. Operations at marine terminals are generally protected by a breakwater. These breakwaters whether are rubble mound type or caissons are expensive as constructed in deep waters in 20 m or greater. The conventional downtime methods result generally in a higher value which subsequently requires a longer breakwater once the downtime criteria is exceeded. The advanced approach for estimating downtime helps to optimize the terminal/breakwater layout and subsequently save on the CAPEX. This approach estimates the downtime for the long term environmental time series using an inhouse Matlab code/program developed using neural network. In order to estimate the downtime, a set of specialized studies are conducted first illustrating the breadth and depth of port engineering. First, the wave climate for a long term time years is established at the project site using Spectral Wave Model MIKE 21 SW. The wave conditions are then transformed to the breakwater leeside (terminal side) using Boussinesq wave model MIKE 21 BW. Dynamic mooring study for the vessels at terminal is carried out using the time domain mooring analysis software MIKE 21 MA. Finally, an inhouse developed Matlab program calculates the downtime for the metocean time series based on the dynamic vessel response of the large set of selected environmental combinations. Up to the author’s knowledge, this is the first published work highlighting limitations of conventional methods and the importance of implementing advanced techniques. This leads to a new thinking of how terminals are to be designed in the future.

This study makes use of the artificial intelligence approaches combined with some nonlinear optimization techniques for optimization of a well-known problem in financial engineering called yield curve. Yield curve estimation plays an important role on making strategic investment decisions. In this paper, we use two well-known parsimonious estimation models, Nelson-Siegel and Extended Nelson-Siegel, for the yield curve estimation. The proposed models of this paper are formulated as continuous nonlinear optimization problems. The resulted models are then solved using some nonlinear optimization and meta-heuristic approaches. The optimization techniques include hybrid GPSO parallel trust region-dog leg, Hybrid GPSO parallel trust region-nearly exact, Hybrid GPSO parallel Levenberg-Marquardt and Hybrid genetic electromagnetism like algorithm. The proposed models of this paper are examined using some real-world data from the bank of England and the results are analyzed.

An integrated team artificial intelligence electromagnetic, T-AI-EM environment is developed to accurately determine the performance characteristics of synchronous reluctance machines (SynRM) with axially laminated anisotropic (ALA) rotor configurations. This T-AI-EM approach is also used to develop an optimum design of SynRM for traction applications. The main objective of this optimization is to minimize the torque ripple while maximizing the output torque to ensure high acceleration and smooth operation. Furthermore, this model will enable the SynRM motor to easily interact in real time with the control unit of the hybrid-electric vehicle, HEV. The T-AI-EM is composed of two main blocks. The first block consists of electromagnetic module utilizing indirectly coupled finite element state space (FE-SS) model. The second consists of an AI based model inspired from team member concept that consists of several adaptive network fuzzy inference systems, ANFISs, supervised by a radial based network, RBN.

Up to now, an indirect generalized method to predict pore pressure under subpressured conditions has not been reported in the literature. In this work, an innovative procedure is presented for estimation of pore pressure and optimization of wells drilled in the abnormally subpressured Deep Basin of the Western Canada Sedimentary Basin (WCSB). The procedure starts with detailed evaluation of 5 wells drilled in a township that covers the study area. Pore Pressure was calculated from Sonic Logs and the Modified D Exponent using Eaton’s Method (Eaton, 1975), which proved to be the most effective approach for abnormal subpressured conditions over a variety of methods tested (Contreras et al., 2011). The optimization procedure was carried out using the Apparent Rock Strength Log (ARSL), which is an effective indicator of formation-drillability and very sensitive to the pore pressure. Next, optimization of individual sections in each well was carried out to determine the optimum types of bits and operational parameters for the lowest cost of drilling. An artificial intelligence function was implemented to set up the optimum combination of parameters in such a way that the rate of penetration (ROP) (m/h) was increased after a number of simulation runs while controlling the bit wear. Especial attention was focused on tight gas reservoirs for selection of the most suitable parameters that increase the quality of drill cuttings. It was concluded that the rollercone bit IADC 547 with at least 0.73 horsepower in the bit per square inch provides the best quality of cuttings for the Nikanassin Group. This is of paramount importance for increasing accuracy in the quantitative determination of permeability and porosity from cuttings particularly in those tight gas reservoirs where the amount of cores is very limited. It is concluded that wells in the Deep Basin of the WCSB can be drilled efficiently with 7 bit runs while keeping the cuttings quality, bit wear and well stability at a significantly high average ROP of 13 m/h. Another conclusion is that the normal trend methods from sonic logs are the most effective approach when dealing with an abnormally subpressured basin.

Reservoirs play an important role within the water resources management framework. Among the available techniques for reservoir optimal operation, the most well known one is stochastic dynamic programming (SDP). In recent years, artificial intelligence techniques such as genetic algorithms (GA) and artificial neural networks have arisen as an alternative to overcome some of the limitations of traditional methods. Some of these limitations are related to the difficulty in combining SDP with other simulation and prediction models, the curse of dimensionality due to the increase in the number of decision and state variables, and the error resulting from the rough discretization of these variables. Here, we introduce a new approach for system optimization and operation, named stochastic fuzzy neural network (SFNN), which can be defined as a neuro-fuzzy system that is stochastically trained (optimized) by a GA model to represent the system operational strategy. Moreover, to deal with imprecision originated by the discretization of inflow intervals (events) in calculating the transition probabilities, we applied the method based on the conditional probability of a fuzzy event. To investigate the applicability and efficiency of the proposed method, the Barra Bonita Reservoir, Brazil, is stochastically optimized and operated. The results found by the SFNN method were compared to the results of other available dynamic programming models, showing success in developing and applying the proposed method to optimal reservoir operation.

Optimization of an adverse outcome pathway network on chemical-induced cholestasis using an artificial intelligence-assisted data collection and confidence level quantification approach