Minimal Computational Complexity Research Articles

The interplay of computational complexity and memory load during quantifier verification

ABSTRACTFormal analysis of the minimal computational complexity of verification algorithms for natural language quantifiers implies that different classes of quantifiers demand the engagement of different cognitive resources for their verification. In particular, sentences containing proportional quantifiers, e.g. “most”, provably require a memory component, whereas non-proportional quantifiers, e.g. “all”, “three”, do not. In an ERP study, we tested whether previously observed differences between these classes were modulated by memory load. Participants performed a picture-sentence verification task while they had to remember a string of 2 or 4 digits to be compared to a second string at the end of a trial. Relative to non-proportional quantifiers, proportional quantifiers elicited a sentence-internal sustained negativity. Additionally, an interaction between Digit-Load and Quantifier-Class was observed at the sentence-final word. Our results suggest that constraints on cognitive resources deployed during human sentence processing and verification are of the same nature as formal constraints on abstract machines.

NeuBTF: Neural fields for BTF encoding and transfer

Neural material representations are becoming a popular way to represent materials for rendering. They are more expressive than analytic models and occupy less memory than tabulated BTFs. However, existing neural materials are immutable, meaning that their output for a certain query of UVs, camera, and light vector is fixed once they are trained. While this is practical when there is no need to edit the material, it can become very limiting when the fragment of the material used for training is too small or not tileable, which frequently happens when the material has been captured with a gonioreflectometer. In this paper, we propose a novel neural material representation which jointly tackles the problems of BTF compression, tiling, and extrapolation. At test time, our method uses a guidance image as input to condition the neural BTF to the structural features of this input image. Then, the neural BTF can be queried as a regular BTF using UVs, camera, and light vectors. Every component in our framework is purposefully designed to maximize BTF encoding quality at minimal parameter count and computational complexity, achieving competitive compression rates compared with previous work. We demonstrate the results of our method on a variety of synthetic and captured materials, showing its generality and capacity to learn to represent many optical properties.

A deep learning framework for early diagnosis of Alzheimer’s disease on MRI images

Numerous medical studies have shown that Alzheimer’s disease (AD) was present decades before the clinical diagnosis of dementia. As a result of the development of these studies with the discovery of many ideal biomarkers of symptoms of Alzheimer’s disease, it became clear that early diagnosis requires a high-performance computational tool to handle such large amounts of data, as early diagnosis of Alzheimer’s disease provides us with a healthy opportunity to benefit from treatment. The main objective of this paper is to establish a complete framework that is based on deep learning approaches and convolutional neural networks (CNN). Four stages of AD, such as (I) preprocessing and data preparation, (II) data augmentation, (III) cross-validation, and (IV) classification and feature extraction based on deep learning for medical image classification, are implemented. In these stages, two methods are implemented. The first method uses a simple CNN architecture. In the second method, the VGG16 model is the pre-trained model that is trained on the ImageNet dataset but applies the same model to the different datasets. We apply transfer learning, meaning, and fine-tuning to take advantage of the pre-trained models. Seven performance metrics are used to evaluate and compare the two methods. Compared to the most recent effort, the proposed method is proficient of analyzing AD, moreover, entails less labeled training samples and minimal domain prior knowledge. A significant performance gain on classification of all diagnosis groups was achieved in our experiments. The experimental findings demonstrate that the suggested designs are appropriate for basic structures with minimal computational complexity, overfitting, memory consumption, and temporal regulation. Besides, they achieve a promising accuracy, 99.95% and 99.99% for the proposed CNN model in the classification of the AD stage. The VGG16 pre-trained model is fine-tuned and achieved an accuracy of 97.44% for AD stage classifications.

Optimal trajectory planning of the industrial robot using hybrid S-curve-PSO approach

PurposeThe paper aims to develop an efficient and compact hybrid S-curve-PSO (particle swarm optimization) controller for the optimal trajectory planning of industrial robots in the presence of obstacles, especially those used in pick-and-place operations.Design/methodology/approachThe proposed methodology comprises a monotonic trajectory through bounded entropy of speed, velocity, acceleration and jerk. Thus, the robot’s trajectory planning corresponds with S-curve-PSO duality. This is achieved by dual navigation with minimal computational complexity. The matrix algebra-based computational complexity transforms the trajectory from random to compact. The linear programming problem represents the proposed robot in Euclidean space, and its optimal solution sets the corresponding optimal trajectory.FindingsThe proposed work ensures the efficient trajectory planning of the industrial robot in the presence of obstacles with optimized path length and time. The real-time and simulation analysis of the robot is presented for performance measurement, and their outcomes demonstrate a good correlation. Compared with the existing controller, it gives a noteworthy improvement in performance.Originality/valueThe novel S-curve-PSO hybrid approach is presented here, along with the LIDAR sensors, which generate the environment map and detect obstacles for autonomous trajectory planning. Based on the sensory information, the proposed approach generates the optimal trajectory by avoiding obstacles and minimizing the travel time, jerk, velocity and acceleration. The hybrid S-curve-PSO approach for optimal trajectory planning of the industrial robot in the presence of obstacles has not been presented by any researchers. This method considers the robot’s kinematics as well as its dynamics. The implementation of the PSO makes it computationally superior and faster. The selection of best-fit parameters by PSO assures the optimized trajectory in the presence of obstacles and uncertainty.

CAP2M..÷Contingent Anonymity Preserving Privacy Method for the Internet of Things Services

The Internet of Things (IoT) consists of a collection of inter-connected devices that are used to transmit data. Secure transactions that guarantee user anonymity and privacy are necessary for the data transmission process. One of the major constraints of IoT Healthcare is the issue of maintaining security. These problems are addressed in this study by the Contingent Anonymity-Preserving Privacy Method (CAP2M), which takes into account the various security needs. The CAP2M method uses the session deployment and user requirements to preserve user privacy. The initial and final privacy settings are modified in the session deployment depending on the session interval and the service provider's recommendation. Federated learning is employed for training privacy settings toward the session's final interval. The suggested approach maintains privacy on various sharing intervals with minimum computation complexity. In the privacy-preserving process, two-factor authentication is used for concealing the sharing and shared ends. The performance is validated using the metrics failure, complexity, and latency.

Generative Adversarial and Dual Layered Deep Classification Techniques for Improving Block Constructions in Public Cloud

This paper provides the Generative Adversarial and Dual Layered Deep Classification techniques to improve the drawbacks in the methods of Absolute Moment BTC (AMBTC) technique in reconstruction error rate of standard BTC model. The image blocks generation and compression are the main phases of BTC model. This can be applied for both colour images and grey scale images. However, the conventional BTC procedures lacks for edge reconstructions and noise reductions in the output images. The first technique GABTC is developed with multi-layered Deep Neural Network (DNN) structures with GA neural models. The integration of both GA models and BTC principles improve the quality of block constructions and reconstructions significantly. The second proposed work is adopted the Dual layered Deep Classification Technique. Handling the image database with minimal storage complexity, minimal computational complexity and optimal quality is a significant task. To obtain these solutions, many image processing techniques are evolved. In the domain, image compression and decompression are more needed at any cost for effectively handling the complex image databases. E-Learning resources are widely used around the internet based knowledge sharing environments. In the E-Learning environment, multiple types of data resources are managed. Particularly, organizing the images is more crucial task where multiple qualities of images are appeared inside the E-Learning network databases. This problem expects solutions from effective image compression techniques. Block Truncation Coding (BTC) and Absolute Moment BTC (AMBTC) are the techniques provide useful and easy implementations of E-Learning based image compression platform. At the same time, they are limited to image dissimilarity rate. To maintain the quality of images in both compression and decompression phases, multilevel image analysis models and training phases are required. In this regard, this proposed system develops a Dual Layered Deep Classification and Truncation (DLDCT) technique. DLDCT comprises the baseline benefits of BTC, multi-layered Support Vector Machine (SVM) units and Deep Layered Convolutional Neural Network (DLCNN) for producing classified range of image pixels and compressing the images under controlled circumstances. This proposed DLDCT makes the image compression and decompression with determined observations. This reduces real time errors occur during image reconstruction phases. This proposed system has been implemented and compared with existing works with respect to significant performance parameters.

RBFK cipher: a randomized butterfly architecture-based lightweight block cipher for IoT devices in the edge computing environment

Internet security has become a major concern with the growing use of the Internet of Things (IoT) and edge computing technologies. Even though data processing is handled by the edge server, sensitive data is generated and stored by the IoT devices, which are subject to attack. Since most IoT devices have limited resources, standard security algorithms such as AES, DES, and RSA hamper their ability to run properly. In this paper, a lightweight symmetric key cipher termed randomized butterfly architecture of fast Fourier transform for key (RBFK) cipher is proposed for resource-constrained IoT devices in the edge computing environment. The butterfly architecture is used in the key scheduling system to produce strong round keys for five rounds of the encryption method. The RBFK cipher has two key sizes: 64 and 128 bits, with a block size of 64 bits. The RBFK ciphers have a larger avalanche effect due to the butterfly architecture ensuring strong security. The proposed cipher satisfies the Shannon characteristics of confusion and diffusion. The memory usage and execution cycle of the RBFK cipher are assessed using the fair evaluation of the lightweight cryptographic systems (FELICS) tool. The proposed ciphers were also implemented using MATLAB 2021a to test key sensitivity by analyzing the histogram, correlation graph, and entropy of encrypted and decrypted images. Since the RBFK ciphers with minimal computational complexity provide better security than recently proposed competing ciphers, these are suitable for IoT devices in an edge computing environment.

Brain Tumor Classification Using Meta-Heuristic Optimized Convolutional Neural Networks

The field of medical image processing plays a significant role in brain tumor classification. The survival rate of patients can be increased by diagnosing the tumor at an early stage. Several automatic systems have been developed to perform the tumor recognition process. However, the existing systems could be more efficient in identifying the exact tumor region and hidden edge details with minimum computation complexity. The Harris Hawks optimized convolution network (HHOCNN) is used in this work to resolve these issues. The brain magnetic resonance (MR) images are pre-processed, and the noisy pixels are eliminated to minimize the false tumor recognition rate. Then, the candidate region process is applied to identify the tumor region. The candidate region method investigates the boundary regions with the help of the line segments concept, which reduces the loss of hidden edge details. Various features are extracted from the segmented region, which is classified by applying a convolutional neural network (CNN). The CNN computes the exact region of the tumor with fault tolerance. The proposed HHOCNN system was implemented using MATLAB, and performance was evaluated using pixel accuracy, error rate, accuracy, specificity, and sensitivity metrics. The nature-inspired Harris Hawks optimization algorithm minimizes the misclassification error rate and improves the overall tumor recognition accuracy to 98% achieved on the Kaggle dataset.

A novel deep learning architecture for disease classification in Arabica coffee plants

SummarySeveral research works on disease detection in coffee plants have been presented in recent years. Leaf miner and rust are the most prevalent diseases in Arabica coffee plants. Early detection of such diseases allows farmer to take diagnostic actions before the infection spreads to neighboring plants. With advancements in drones and artificial intelligence (AI), the automatic detection of leaf diseases is gaining prominence in the field of smart agriculture. Furthermore, it is critical to develop an accurate method for infestation detection with minimal computational complexity. Existing works for plant disease detection utilize pre‐trained deep learning models with millions of parameters. A feasible trade‐off has to be attained between accuracy and computational complexity for the deployment of such deep networks. This research proposes an effective method for disease detection in Arabica coffee plants using EfficientNetB0 architecture. The architecture of the EfficientNetB0 network was improvised by including a ghost module at its end. This integration allows the network to learn effectively with minimal parameters without compensating for the end accuracy. The proposed model has a total of 4,874,531 parameters which is significantly lesser than most of the state‐of‐the‐art deep learning architectures and achieved an accuracy of 84%.

Modeling the impact of the Omicron infection wave in Germany.

In November 2021, the first infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variant of concern (VOC) B.1.1.529 ('Omicron') was reported in Germany, alongside global reports of reduced vaccine efficacy (VE) against infections with this variant. The potential threat posed by its rapid spread in Germany was, at the time, difficult to predict. We developed a variant-dependent population-averaged susceptible-exposed-infected-recovered infectious-disease model that included information about variant-specific and waning VEs based on empirical data available at the time. Compared to other approaches, our method aimed for minimal structural and computational complexity and therefore enabled us to respond to changes in the situation in a more agile manner while still being able to analyze the potential influence of (non-)pharmaceutical interventions (NPIs) on the emerging crisis. Thus, the model allowed us to estimate potential courses of upcoming infection waves in Germany, focusing on the corresponding burden on intensive care units (ICUs), the efficacy of contact reduction strategies, and the success of the booster vaccine rollout campaign. We expected a large cumulative number of infections with the VOC Omicron in Germany with ICU occupancy likely remaining below capacity, nevertheless, even without additional NPIs. The projected figures were in line with the actual Omicron waves that were subsequently observed in Germany with respective peaks occurring in mid-February and mid-March. Most surprisingly, our model showed that early, strict, and short contact reductions could have led to a strong 'rebound' effect with high incidences after the end of the respective NPIs, despite a potentially successful booster campaign. The results presented here informed legislation in Germany. The methodology developed in this study might be used to estimate the impact of future waves of COVID-19 or other infectious diseases.

Load-Aware VM Migration Using Hypergraph Based CDB-LSTM

Live Virtual Machine (VM) migration is one of the foremost techniques for progressing Cloud Data Centers’ (CDC) proficiency as it leads to better resource usage. The workload of CDC is often dynamic in nature, it is better to envisage the upcoming workload for early detection of overload status, underload status and to trigger the migration at an appropriate point wherein enough number of resources are available. Though various statistical and machine learning approaches are widely applied for resource usage prediction, they often failed to handle the increase of non-linear CDC data. To overcome this issue, a novel Hypergrah based Convolutional Deep Bi-Directional-Long Short Term Memory (CDB-LSTM) model is proposed. The CDB-LSTM adopts Helly property of Hypergraph and Savitzky–Golay (SG) filter to select informative samples and exclude noisy inference & outliers. The proposed approach optimizes resource usage prediction and reduces the number of migrations with minimal computational complexity during live VM migration. Further, the proposed prediction approach implements the correlation co-efficient measure to select the appropriate destination server for VM migration. A Hypergraph based CDB-LSTM was validated using Google cluster dataset and compared with state-of-the-art approaches in terms of various evaluation metrics.

Intelligent Intrusion Detection System for Industrial Internet of Things Environment

Rapid increase in the large quantity of industrial data, Industry 4.0/5.0 poses several challenging issues such as heterogeneous data generation, data sensing and collection, real-time data processing, and high request arrival rates. The classical intrusion detection system (IDS) is not a practical solution to the Industry 4.0 environment owing to the resource limitations and complexity. To resolve these issues, this paper designs a new Chaotic Cuckoo Search Optimization Algorithm (CCSOA) with optimal wavelet kernel extreme learning machine (OWKELM) named CCSOA-OWKELM technique for IDS on the Industry 4.0 platform. The CCSOA-OWKELM technique focuses on the design of feature selection with classification approach to achieve minimum computation complexity and maximum detection accuracy. The CCSOA-OWKELM technique involves the design of CCSOA based feature selection technique, which incorporates the concepts of chaotic maps with CSOA. Besides, the OWKELM technique is applied for the intrusion detection and classification process. In addition, the OWKELM technique is derived by the hyperparameter tuning of the WKELM technique by the use of sunflower optimization (SFO) algorithm. The utilization of CCSOA for feature subset selection and SFO algorithm based hyperparameter tuning leads to better performance. In order to guarantee the supreme performance of the CCSOA-OWKELM technique, a wide range of experiments take place on two benchmark datasets and the experimental outcomes demonstrate the promising performance of the CCSOA-OWKELM technique over the recent state of art techniques.

Brain tumor segmentation using a deep Shuffled‐YOLO network

AbstractBrain tumor is an anomalous proliferation of cells in the brain that can evolve to malignant and benign tumors. Currently, segmentation of brain tumor is the most important surgical and pharmaceutical procedures. However, manually segmenting brain tumors is hard because it is hard to find erratically shaped tumors with only one modality; the MRI modalities are integrated to provide multi‐modal images with data that can be utilized to segment tumors. The recent developments in machine learning and the accessibility of medical diagnostic imaging have made it possible to tackle the challenges of segmenting brain tumors with deep neural networks. In this work, a novel Shuffled‐YOLO network has been proposed for segmenting brain tumors from multimodal MRI images. Initially, the scalable range‐based adaptive bilateral filer (SCRAB) pre‐processing technique was used to eliminate the noise artifacts from MRI while preserving the edges. In the segmentation phase, we propose a novel deep Shuffled‐YOLO architecture for segmenting the internal tumor structures that include non‐enhancing, edema, necrosis, and enhancing tumors from the multi‐modality MRI sequences. The experimental fallouts reveal that the proposed Shuffled‐YOLO network achieves a better accuracy range of 98.07% for BraTS 2020 and 97.04% for BraTS 2019 with very minimal computational complexity compared to the state‐of‐the‐art models.

On the Performance of Rate-Compatible LDPC Codes under All-Integer Quantization Decoding

From a practical point of view, quantization is necessary for the design of any low-complexity LDPC decoder. However, the quantization scheme with the minimum computational complexity called all-integer quantization has not been well investigated since, intuitively, it seems to significantly degrade the decoding performance. It is found in this work that the utilization of 4-bit all-integer quantization for MSA-based LDPC decoder is adequate to achieve the decoding performance of the nonquantized case which can be considered as the lower bound on the decoding performance. Moreover, this simple quantization scheme can also be applied to the hardware and cost-efficient rate-compatible LDPC codes.

LssDet: A Lightweight Deep Learning Detector for SAR Ship Detection in High-Resolution SAR Images

Synthetic aperture radar (SAR) ship detection has been the focus of many previous studies. Traditional SAR ship detectors face challenges in complex environments due to the limitations of manual feature extraction. With the rise of deep learning (DL) techniques, SAR ship detection based on convolutional neural networks (CNNs) has achieved significant achievements. However, research on CNN-based SAR ship detection has mainly focused on improving detection accuracy, and relatively little research has been conducted on reducing computational complexity. Therefore, this paper proposes a lightweight detector, LssDet, for SAR ship detection. LssDet uses Shufflenet v2, YOLOX PAFPN and YOLOX Decopuled Head as the baseline networks, improving based on the cross sidelobe attention (CSAT) module, the lightweight path aggregation feature pyramid network (L-PAFPN) module and the Focus module. Specifically, the CSAT module is an attention mechanism that enhances the model’s attention to the cross sidelobe region and models the long-range dependence between the channel and spatial information. The L-PAFPN module is a lightweight feature fusion network that achieves excellent performance with little computational effort and a low parametric count. The Focus module is a low-loss feature extraction structure. Experiments showed that on the Sar ship detection dataset(SSDD), LssDet’s computational cost was 2.60 GFlops, the model’s volume was 2.25 M and AP@[0.5:0.95] was 68.1%. On the Large-scale SAR ship detection dataset-v1.0 (LS-SSDD-v1.0), LssDet’s computational cost was 4.49 GFlops, the model’s volume was 2.25 M and AP@[0.5:0.95] was 27.8%. Compared to the baseline network, LssDet had a 3.6% improvement in AP@[0.5:0.95] on the SSDD, and LssDet had a 1.5% improvement in AP@[0.5:0.95] on the LS-SSDD-v1.0. At the same time, LssDet reduced Floating-point operations per second (Flops) by 7.1% and Paraments (Params) by 23.2%. Extensive experiments showed that LssDet achieves excellent detection results with minimal computational complexity. Furthermore, we investigated the effectiveness of the proposed module through ablation experiments.

Improve cardinality with two-dimensional unipolar (optical) orthogonal codes for multiple access interference

Abstract Optical code division multiple access (OCDMA) is one of the promising technologies to be implemented using all-optical networks. The OCDMA system performance is substantially determined by the particular encoding codes. Due to the different transmitter and receiver design considerations, selecting the right code family and code dimension is critical. In this research, 2D codes based on permutation vectors’ (PV) codes are presented to overcome the Multiple Access Interference (MAI) problem and system complexity. Consequently, the research work focused on two-dimensional unipolar codes to mitigate the probability error and cardinality of the optical CDMA system. A fast wavelength-hopping/time-spreading (FWHTS) is introduced as an encoding technique. The FWHTS is used for code construction design, and it achieves cardinality in upper bound theory without sacrificing good auto-and-cross correlation properties. Accordingly, hybrid codes with serial elimination interference receivers are presented to estimate the error probability of 2D optical codes. The proposed method is simulated using Python software. The performance metrics are flexibility, bit rate, minimum temporal length, computational complexity, cardinality and wavelength. The proposed method is compared with the existing 2D balanced incomplete block design (BIBD) technique and multidiagonal (MD) codes. The performance of the proposed method is approximately 4% higher than these existing methods. Subsequently, in the future work, a SAC-OCDMA code construction design shall be presented and the decoding techniques can be introduced, which can reduce the receiver noise and other phase intensity induced noise, respectively.

Performance Improvement of a Parsimonious Learning Machine Using Metaheuristic Approaches.

Autonomous learning algorithms operate in an online fashion in dealing with data stream mining, where minimum computational complexity is a desirable feature. For such applications, parsimonious learning machines (PALMs) are suitable candidates due to their structural simplicity. However, these parsimonious algorithms depend upon predefined thresholds to adjust their structures in terms of adding or deleting rules. Besides, another adjustable parameter of PALM is the fuzziness in membership grades. The best set of such hyper parameters is determined by experts' knowledge or by optimization techniques such as greedy algorithms. To mitigate such experts' dependency or usage of computationally expensive greedy algorithms, in this work, a meta heuristic-based optimization technique, called the multimethod-based optimization technique (MOT), is utilized to develop an advanced PALM. The performance has been compared with some popular optimization techniques, namely, the greedy search, local search, genetic algorithm (GA), and particle swarm optimization (PSO). The proposed parsimonious learning algorithm with MOT outperforms the others in most cases. It validates the multioperator-based optimization technique's advantages over the single operator-based variants in selecting the best feasible hyperparameters for the autonomous learning algorithm by maintaining a compact architecture.

Compression of Cyber Learning Images Based on DLDCT

Dealing with the picture database with minimal garage complexity, minimum computational complexity and ultimate nice is an important work. To obtain these solutions, many image-processing techniques are advanced. Now days, E-learning assets are widely used across the internet based totally expertise sharing environments. Within the cyber learning surroundings, multiple sorts of records sources are controlled. Particularly, organizing the photos is challenge that is more vital where pictures appeared in the cyber learning network databases. This trouble expects solutions from powerful image compression techniques. Block truncation coding technique is offer beneficial and easy implementations of cyber learning to know primarily based picture compression platform. On this regard, this proposed system develops a Dual Layered Deep Classification and Truncation (DLDCT) approach. This proposed system has applied the DLDCT and as compared with existing works with admire to considerable overall performance parameters.

Factorization of the Degree of Sphenic Polynomials Over the Galois Fields of Arbitrary Characteristics

By sphenic polynomials, we mean polynomials formed by the product of three (not necessarily different) irreducible polynomials with a priori unknown degree. The study's main goal is to develop an effective algorithm for factorizing degrees of sphenic polynomials with minimal computational complexity. Different solutions to the problem of factorization degrees of sphenic polynomials depending on the ratio degree of the cycle period of these polynomials consider. The sphenic polynomial cycle period defines as a parameter equal to the number of non-repeating subtractions computed on the linear-logarithmic scale of the group formed by the sphenic polynomial. The proposed algorithm is invariant to the characteristics of the Galois fields generated by the multipliers of sphenic polynomials. Numerous numerical examples confirm the correctness of the results. Directions for further research outlines.

Intelligent conventional and proposed hybrid 5G detection techniques

In recent years, Multiple Inputs and Multiple Outputs (MIMO) have gained significant attention due to their characteristics such as spectral gain, high throughput, and energy efficient. It is seen as one of the integral parts and backbone of the Fifth Generation (5G) and beyond 5G (B5G). However, the use of a large number of antennas requires complex algorithms to detect the received signal. Though, several detection methods have been proposed which can efficiently enhance the Bit Error Rate (BER) gain of the framework, it also increases the computational complexity. The proposed article introduces a hybrid algorithm for different sizes of MIMO. The hybrid algorithm is designed by combining QR Decomposition M−algorithm−Maximum Likelihood Detection (QRM-MLD) and Beam Forming (BF). Further, we compare the performance of proposed hybrid algorithms with that of conventional algorithms, namely Zero Forcing (ZF), Minimum Mean Square Error (MMSE), Successive over Relaxation (SOR), Gauss Seidel Detector (GSD), Jacobi Scheme (JS), and Approximate Message Passing (AMP). In computer simulation, it is noted that the proposed algorithm outperforms the conventional detection algorithms with minimum computational complexity.