Hepatocellular Carcinoma (HCC) is a type of liver cancer that contributes for about 75

Video memorability prediction is a challenging task with applications in numerous domains, such as advertising, social media, and education. This paper proposes a multimodal deep learning approach to video memorability prediction that leverages both visual and textual information. A novel language-guided frame extraction process is also proposed to extract representative frames from the video. The proposed architecture, MemXNet, leverages spatiotemporal features from input video, semantic embeddings from the text description of the video, and visual features from the representative frames of the video to obtain a holistic representation for video memorability prediction. The performance of the model is evaluated on the VideoMem dataset and the results are presented. In-depth ablation studies are conducted to examine the contribution of each modality to the overall memorability score. This study contributes to the growing body of research on multimodal deep learning for video analysis and paves the way for future research in this area.

Dentistry diseases are worldwide concern with the fact that 5% of medical budget is spent on it. Dental diagnosis requires frequent visits to clinic and multiple personal check‐ups by an expert. This delays the process of diagnosis as well as introduces the danger of oral infection spread. Automation is boon to speed up the process of diagnosis, reduce expert's involvement and help in handling volume of patients. In situations like pandemic as seen in this decade, computer based automatic systems in health have proven their importance and necessity. For dental diagnosis, images are useful tool for better anatomical views and accurate decision of treatment. However, manual dental analysis increases load on dentists for initial check‐up that can be easily performed with automated systems or self‐kits. Oral cavity is a dental illness, if not identified and treated on time may lead to other serious ailments. This work presents a framework that performs tooth image classification for cavity and non‐cavity with new bag of features (NBoF) method. NBoF method is an attempt to improve the performance of bag of features (BoF) using the proposed reinforcement Aquila optimization (RAO) and weighted Bayesian Gaussian mixture modelling (WBGMM). An analysis of the performance of the NBoF using the proposed RAO and WBGMM is conducted using standard metrics. The comparative study of results proves that the proposed NBoF method outperforms the existing state‐of‐the‐art algorithms.

In the current age of widespread application of artificial intelligence (AI) across various facets of life, satellite remote sensing is no outlier. Thanks to the ongoing enhancements in the spatial and temporal resolutions of satellite images, they are emerging as invaluable assets in areas such as land-use analysis, meteorology, change detection, and beyond. Accurate analysis and classification at various levels of hyperspectral images (HSIs) and multispectral remote sensing images (RSIs) are essential for extracting valuable insights from these datasets.

Music genres play a pivotal role in structuring and interpreting music through shared attributes and traditions. Accurate classification of music genres enhances the overall music experience and enables more precise evaluation of artistic innovation and quality. In this paper, a music genre classification system has been proposed that uses an Artificial Neural Network (ANN) Model, moving beyond the traditional Convolutional Neural Network (CNN) approaches that rely heavily on spectrogram images and require extensive computational resources. Our model focuses on analyzing only the crucial characteristic features of audio files, sidestepping the need for complete spectrogram data. Mel-frequency Cepstral Coefficient (MFCC) has been utilized as the main feature extraction method and is combined with Chroma features and Spectral Contrast. The widely used GTZAN dataset has been utilized. After the extraction of the feature vector, the proposed model classifies the music into different genres. The model exhibited impressive performance with a training accuracy of 98.41% and a validation accuracy of 97.05%, highlighting its effectiveness in accurately classifying music genres.

The pervasive influence of social media necessitateseffective strategies to combat the rapid dissemination of misinformation. This paper delves into the intricate task of rumor detection on Twitter, employing a multifaceted approach. The authorsleverage the capabilities of classical machine learning models, like SVM and Random Forests, to establish a baseline. Expanding the scope, the authors explore the potential of Artificial Neural Networks, leveraging their capacity to capture nuanced patterns in data. Additionally, they employ advanced transformer-based models like BERT and RoBERTa to enhance the understanding of semantic context. The study encompasses thorough experimentation with various text vectorization techniques, dimensionality reduction methods such as LDA, and comprehensive model evaluations. The findings reveal the supremacy of the BERT model, achieving an accuracy of 90.05%. This comprehensive analysis contributes valuable insights to the ongoing discourse on mitigating the impact of misinformation in the realm of social media.

In modern society numerous digital devices play a very significant role in day-to-day life. Digital devices are well connected and easily accessible through multiple sensors and Internet of Things (IoT) devices. Due to the rapid growth of digital devices, large amount of data traffics are being generated, which induces network congestion. To deal with large amount of data traffic a programmable Software Defined IoT (SD-IoT) infrastructure is utilized. For efficient and sustainable network, the data must be transmitted through optimal path in such a way to as to minimize energy consumption. Here, the network is partitioned into clusters to find an optimal path. Finding an optimal path from a set of possible paths is an NP-complete problem. To solve this problem, we propose to find a set of optimal border nodes of each cluster with other clusters in the network, so as to reduce the number of possible paths between clusters. The set of optimal border nodes will be selected in such a way so that they have maximum energy and minimum distances. This paper proposes an intelligent approach to find the set of optimal border nodes using Lion Swarm Optimization algorithm (LSOA). Once a set of optimal border nodes are obtained, an optimal path can be generated using a routing mechanism. The performance of the proposed work is analyzed in terms of packet delivery ratio, average latency, network lifetime and energy consumptions. The results show that the border nodes selected using LSOA finds better routes as compared to the border nodes selected using other state-of-the-art metaheuristics algorithm thereby, increases suitability of the network by energy conservation.

Osteosarcoma is one of the most common malignant bone tumors mostly found in children and teenagers. Manual detection of osteosarcoma requires expertise and it is a labour-intensive process. If detected on time, the mortality rate can be reduced. With the advent of new technologies, automatic detection systems are used to analyse and classify medical images, which reduces the dependency on experts and leads to faster processing. In this paper, an automatic detection system: Integrated Features-Feature Selection Model for Classification (IF-FSM-C) to detect osteosarcoma from the high-resolution whole slide images (WSIs) is proposed. The novelty of the proposed approach is the use of integrated features obtained by fusion of features extracted using traditional handcrafted (HC) feature extraction techniques and deep learning models (DLMs) namely EfficientNet-B0 and Xception. To further improve the performance of the proposed system, feature selection (FS) is performed. Here, two binary variants of recently proposed Arithmetic Optimization Algorithm (AOA) known as BAOA-S and BAOA-V are proposed to perform FS. The selected features are given to a classifier that classifies the WSIs into Viable tumor (VT), Non-viable tumor (NVT) and non-tumor (NT). Experiments are performed to compare the performance of proposed IF-FSM-C to the classifiers which use HC or deep learning features alone as well as state-of-the-art methods for osteosarcoma detection. The best overall accuracy of 96.08% is obtained when integrated features extracted using HC techniques and Xception are used. The overall accuracy is enhanced to 99.54% after applying BAOA-S for FS. Further, the application of BAOA-S for FS reduces the number of features with the best model having only 188 features compared to 2118 features if no FS is applied.

The automatic recognition and classification of infected plant leaves play an important role in precision agriculture and in helping to improve crop yields. With the advancements in the fields of artificial intelligence and computer vision, an exponential progress has been observed in their applications to agriculture, such as in plant leaf disease detection and subsequent decision making. However, the complexity and diversity in the structural background of plant leaf images pose several challenges. This article introduces a novel bag-of-features algorithm, called Upgraded Slime mould Algorithm-Bag of Features (USMA-BOF) , for an effective classification of plant leaf images. The stochastic-based upgraded USMA is proposed to determine the optimal visual words. Further, the occurrence or repetition of optimal visual features is represented through a new discrete dual complex chirplet transform (DDCCT) method. Finally, the classification is performed using two classifiers: a support vector machine (SVM) and a multilayer perceptron (MLP). The performance of USMA is first compared with state-of-the-art algorithms on IEEE Congress on Evolutionary Computation (CEC) 2017 benchmark functions. Then, the proposed USMA-BOF algorithm is applied on a classification dataset with binary outputs for disease identification in sustainable agriculture. The MLP classifier performs better than SVM, with an average accuracy of 0.7552 using the proposed USMA-BOF algorithm, as compared to 0.7262, 0.6989, 0.6343, and 0.7262 using the algorithms whale optimization algorithm (WOA)-BOF, adaptive particle swarm optimization (APSO)-BOF, gray wolf optimization (GWO)-BOF, and SMA-BOF, respectively.

Out of the various types of primary bone cancers, Osteosarcoma is one of the most common malignant bone tumors. Children and teenagers are mostly affected by Osteosarcoma, which weakens the strength of their bones and sometimes may even result in death. It is important to develop an intelligent classifier that detect osteosarcoma accurately so that proper treatment can be given to the patient timely. In this paper, we propose an intelligent classifier that classifies osteosarcoma whole slide images (WSIs) into Viable Tumor, Non-Viable Tumor and Non-Tumor. To extract the region of interest (ROI) from WSIs, a Multi-Feature Non-Seed-based Region Growing algorithm (MFNSRG) based on intra-region homogeneity and inter-region heterogeneity maximization is used. We use textural heterogeneity along with color heterogeneity as the matching criteria during region growing. Finally, the background is eliminated using thresholding based on the size of a region and the ROI is obtained. The performance of MFNSRG is further improved by using Marine Predators, a recently proposed metaheuristic algorithm, which is used to obtain optimal value of segmentation parameters. Here, we use handcrafted methods to extract relevant features from the segmented image which are given as input to the classifier. The results of experimentation prove the superiority of the proposed approach as compared to the existing state-of-the-art algorithms.

Out of the various types of skin cancers, melanoma is observed to be the most malignant and fatal type. Early detection of melanoma increases the chances of survival which necessitates the need to develop an intelligent classifier that classifies the dermoscopic images accurately as melanoma or non-melanoma. Features extracted from an image have a major impact on the performance of a classifier. In this paper, handcrafted feature extraction techniques are used to extract features from dermoscopic images. It is quite possible that not all the features extracted from dermoscopic images contribute in the process of classification which imposes the need of selection of significant features from the feature set. Here, two binary variants of Harris Hawk Optimization (HHO) algorithm namely BHHO-S and BHHO-V are presented that employ S-shaped and V-shaped transfer functions with time-dependent behavior, respectively, for feature selection. The selected features are given to a classifier that classifies the dermoscopic image as melanoma or non-melanoma. Comparison of the performance of the proposed methods is done with existing metaheuristic algorithms. The results obtained after experimentation show the superiority of classifier that uses features selected using BHHO-S over BHHO-V and the classifiers that use existing state-of-the-art metaheuristic algorithms. The experimental results also reveal that texture features extracted using local binary pattern along with color features provides higher classification accuracy as compared to global and other local texture feature extraction techniques.

There is an increasing need to get updated information regarding the changes on earth’s surface. The information obtained can be used in a wide range of applications including disaster management, land-use investigation etc. The high-resolution remote sensing images obtained from satellites provide us with an opportunity to detect changes on earth’s surface between various time intervals. In this paper, an unsupervised object-based change detection (OBCD) method is proposed to detect changes in high resolution bi-temporal satellite images. To detect changes, a novel multi-feature non-seed-based region growing (MF-NSRG) algorithm is proposed for image segmentation based on heterogeneity minimization that uses textural heterogeneity along with spectral and spatial heterogeneity during region growing. The performance of MF-NSRG algorithm is further improved by using Harris Hawk, a recently proposed metaheuristic algorithm, which is used to obtain optimal values of segmentation parameters. Finally, the feature maps extracted from the pre-change and post-change segmented images are analysed using histogram trend similarity (HTS) approach to detect changes. The proposed approach is known as object-based change detection using Harris Hawk (OBCD-HH). The proposed OBCD-HH approach is applied on two datasets: xBD and Onera Satellite Change Detection (OSCD) dataset. Its performance is compared with existing state-of-the-art algorithms and results show the superiority of the proposed approach.

Melanoma is amongst the most aggressive form of skin cancer. The manual detection of melanoma using dermoscopic images is a labor-intensive and time-consuming process, which requires a lot of expertise too. This necessitates the development of an intelligent classifier that detects melanoma accurately so that proper and timely treatment can be given to the patient. However, melanoma detection is a challenging task due to the presence of noise like air bubbles, hair, etc. in the dermoscopic images. First, we propose three methods to remove hair in dermoscopic images using various morphological operations. As the quality of features affects the performance of a classifier, we propose to use integrated features that are extracted using handcrafted (HC) feature extraction techniques and deep learning model (DLM) from dermoscopic images, to enhance the performance of the classifier. Two DLMs: ResNet50V2 and EfficientNet-B0 are employed for feature extraction and Artificial Neural Network (ANN) is used for classification. The proposed approach is evaluated using two datasets: the HAM10000 dataset taken from the ISIC 2018 challenge, consisting of 10,015 dermoscopic images belonging to 7 classes, and the PH2 dataset consisting of 200 dermoscopic images with 40 melanoma and 160 non-melanoma images. Experimental results show that the proposed hair removal methods along with the integrated features improve the accuracy of melanoma detection as compared to when (i) no-pre-processing is applied or (ii) HC or deep learning (DL) features are used alone. An accuracy of 94.9% and 98% is achieved on HAM10000 and PH2 datasets respectively when integrated features extracted using HC techniques and EfficientNet-B0 are used for classification along with hair removal techniques. The performance of the proposed approach on both datasets is comparable to the existing state-of-the-art classifiers for melanoma detection.

With the advancement in modern imaging techniques like CT scan, MRI, PET scan etc., a vast amount of data is generated every day in the field of healthcare. Big data contains hidden information, which necessitates the development of intelligent systems to analyze it and extract relevant information, allowing for accurate and cost-effective decisions in the medical field. By utilizing the untapped potential of the big data available in the medical field, very precise models can be developed for the medical diagnosis of retinal diseases. Optical coherence tomography (OCT) is a non-invasive imaging test that captures different, distinctive layers of the retina and optic nerve in a living eye to map and measure their thickness, that helps diagnose various retinal disorders. With the advancement of the application of deep learning-based techniques in the field of medical sciences, the use of convolutional neural network (CNN) based approaches for disease detection is gaining popularity. While the manual examination of 3D OCT images for the diagnosis of retinal disorders requires extensive time and expert intervention, the use of CNNs provides an effective automated option that provides results with higher accuracy while also reducing the time involved in the overall process. In this paper, we have implemented the aforementioned idea by proposing OCT-CNN, a CNN architecture, that automatically classifies retinal OCT images and identifies potential disorders in a living eye. Several techniques have been employed to enhance the performance of the proposed approach, including digital enhancement of the images, dropout regularization, adaptive learning rates, and early stopping of training to attain optimal performance. The performance of the proposed OCT-CNN is evaluated on the UCSD dataset against several popular deep CNN architectures and existing state-of-the-art approaches to automatic retinal OCT classification. The proposed OCT-CNN attains the best performance on all evaluated metrics, pushing the classification accuracies to 99.28

The task of designing an Artificial Neural Network (ANN) can be thought of as an optimization problem that involves many parameters whose optimal value needs to be computed in order to improve the classification accuracy of an ANN. Two of the major parameters that need to be determined during the design of an ANN are weights and biases. Various gradient-based optimization algorithms have been proposed by researchers in the past to generate an optimal set of weights and biases. However, due to the tendency of gradient-based algorithms to get trapped in local minima, researchers have started exploring metaheuristic algorithms as an alternative to the conventional techniques. In this paper, we propose the GGA-MLP (Greedy Genetic Algorithm-Multilayer Perceptron) approach, a learning algorithm, to generate an optimal set of weights and biases in multilayer perceptron (MLP) using a greedy genetic algorithm. The proposed approach increases the performance of the traditional genetic algorithm (GA) by using a greedy approach to generate the initial population as well as to perform crossover and mutation. To evaluate the performance of GGA-MLP in classifying nonlinear input patterns, we perform experiments on datasets of varying complexities taken from the University of California, Irvine (UCI) repository. The experimental results of GGA-MLP are compared with the existing state-of-the-art techniques in terms of classification accuracy. The results show that the performance of GGA-MLP is better than or comparable to the existing state-of-the-art techniques.

The deadliest form of skin cancer is melanoma, and if detected in time, it is curable. Detection of melanoma using biopsy is a painful and time-consuming task. Alternate means are being used by medical experts to diagnose melanoma by extracting features from skin lesion images. Medical image diagnosis requires intelligent systems. Many intelligent systems based on image processing and machine learning have been proposed by researchers in the past to detect different kinds of diseases that are successfully used by healthcare organisations worldwide. Intelligent systems to detect melanoma from skin lesion images are also evolving with the aim of improving the accuracy of melanoma detection. Feature extraction plays a critical role. In this paper, a model is proposed in which features are extracted using convolutional neural network (CNN) with transfer learning and a hierarchical classifier consisting of random forest (RF), k-nearest neighbor (KNN), and adaboost is used to detect melanoma using the extracted features. Experimental results show the effectiveness of the proposed model.

In reality videography possesses complicated dynamics; and open-domain interpretation methods should be time-sensitive and enable both variable-size input (frame sequence) and output (word sequence). To address the issue, here we suggest a pioneering head-to-head or sequence-to-sequence model for creating video captions or descriptions. We use recurrent neural networks, particularly Long Short-Term Memory (LSTM) networks, to do this, since they have shown avant-garde success in the production of picture captions. The training of the LSTM model is based on video-phrase pairs and it acquires its knowledge by equating a series of visual frames with a succession of words to produce a summary of the snippet's occurrence. The proposed model V2CT has shown signs to grasp both the temporal structure of a series of frames and the series model of the formed phrases, i.e., a dialect model. The model is trained and tested on a common collection of YouTube videos (MSVD dataset).

The classification accuracy of a multi-layer perceptron (MLP) depends on the selection of relevant features from the data set, its architecture, connection weights and the transfer functions. Generating an optimal value of all these parameters together is a complex task. Metaheuristic algorithms are popular choice among researchers to solve complex optimization problems. This paper presents a hybrid metaheuristic algorithm simple matching-grasshopper new cat swarm optimization algorithm (SM-GNCSOA) that optimizes all the four components simultaneously. SM-GNCSOA uses grasshopper optimization algorithm, a new variant of binary grasshopper optimization algorithm called simple matching-binary grasshopper optimization algorithm and a new variant of cat swarm optimization algorithm called new cat swarm optimization algorithm to generate an optimal MLP. Features play a vital role in determining the classification accuracy of a classifier. Here, we propose a new feature penalty function and use it in SM-GNCSOA to prevent underfitting or overfitting due to the selected number of features. To evaluate the performance of SM-GNCSOA, different variants of SM-GNCSOA are proposed and their classification accuracies are compared with SM-GNCSOA on ten classification data sets. The results show that SM-GNCSOA gives better results on most of the data sets due to its capability to balance exploration and exploitation and to avoid local minima.