International Journal on Artificial Intelligence Tools

Data augmentation in Natural Language Processing (NLP) faces various challenges that hinder its widespread adoption, unlike its ever-present usage in the field of vision. It is even more the case for the text summarization task where one should focus on both article and summary. In this paper, we review the effect of back translation augmentation, present the diverse beam search decoding strategy, and masking as a method to generate synthetic data for text summarization. The approaches will be evaluated by ROUGE score, novelty, summary length, and GPT-4 to analyze their effectiveness. Our proposed framework presents multiple combinations of back translation and masking for articles, along with diverse augmentation for summaries. Although applicable to networks of any size, we decided to use BART-large, a relatively smaller model, in order to conduct a larger number of experiments. The experiments demonstrated superior performance across all specified metrics when compared to fine-tuning BART-large on the CNN/Dailymail dataset. Specifically, we showed a significant improvement in novelty; 158% and 56% increase rate for bigrams and unigrams, respectively. It could eliminate some copyright concerns around generating content similar to human writing. Additionally, the GPT-4 assessment indicates that models trained using the augmentation technique tend to capture important information more effectively than the baseline model.

Numerous real-world applications, such as online gaming, video streaming, and internet calls are streamed enormous volumes of data. So it is important to quickly process data streams in real-time. Data clustering methods are historically effective and efficient in extracting data from large datasets. Typically, they are ineffective for online data stream clustering. Therefore, an efficient online big data stream clustering using dual interactive Wasserstein generative adversarial network (OBDSC-DI-WGAN) is proposed in this paper. The proposed method consists of three phases: data initialization, online clustering, offline clustering. Initially, the input data are taken from Forest Cover Type dataset. During initialization phase, the dimensions of the input data can be reduced using kernel co-relation approach. After the initialization, the dimension-reduced data are fed to the dual interactive Wasserstein generative adversarial network (DI-WGAN) to accomplish efficient data stream clustering. Then the data enter the selected grid during the stage of online clustering. Afterward, the data stream is activated through the stage of online clustering and the data are activated in the stage of offline depending upon user request. The grid is regarded as a virtual data point in its geometric center during the offline phase. The density radius along cluster centers is determined under Billiards-inspired optimization algorithm. Finally, the clustering outcome is derived from optimum density radius. The proposed technique is activated in MATLAB, and its efficiency is analyzed under some performance metrics, such as accuracy, dice coefficient, purity, sensitivity, specificity, precision, processing time and jacquard coefficient. The proposed method provides better accuracy 27.5%, 10.32% and 16.65%, better precision 30.93%, 11.14% and 15.3% compared with existing methods, like fast grid-based clustering approach for hybrid data stream (FGCH-CCFD-OBDSC), optimized deep autoencoder including CNN for non-stationary environments surveillance data streams (DAE-CNN-OBDSC) and asynchronous dual-pipeline deep learning framework for online data stream classification (1D-CNN-OBDSC) respectively.

Video shot boundary detection (VSBD) plays a key role in analyzing, summarizing, indexing and retrieving content-based data from videos. Many artificial intelligence (AI)-based techniques have recently been introduced to detect the gradual transition from video frames. However, those techniques fail to detect the presence of inter-classes like fade-in, fade-out and dissolve from gradual transition video frames. In addition, the existing techniques face high computational complexity during detecting transitions from the video frames. This research brings novel clustering-based techniques to classify the inter-classes like fade-in, fade-out and dissolve from gradual transition video frames. At the initial stage, color histogram differences (CHD) technique is introduced to detect the abrupt transition from the video frames. The identified abrupt transitions are completely removed from the video frames. Then, segmentation is done to segment the gradual transitions from the video frames. The segmented gradual transition frames are then given to extract the handcrafted and deep features from the video frames. The deep features are extracted from the segments using the Morlet Wavelet-assisted modified stacked autoencoder (MW-MSAE) technique. The extracted handcrafted features and deep features are then concatenated together, and finally, fused feature vectors are obtained. The fused features are then fed to the robust deep $k$-means map clustering method (RDKMM) to aggregate based on similar features. For calculating the similar features, a similarity-based correlation calculation (SBCC) is done in adjacent frames to determine gradual shot transitions like fade-in, fade-out and dissolve. The dataset used in this research is the TREC Video Retrieval Evaluation (TRECVID) 2021 dataset. In the experimental scenario, an accuracy of 95.5%, a sensitivity of 95.3%, a specificity of 96.9%, a precision of 93.9%, an F-measure of 94.6% and Mathew’s correlation coefficient (MCC) of 92.4% are obtained.

In modern times, system energy load forecasting is an extremely important process in a variety of contexts. Moreover, energy load time series fluctuations are influenced by a wide range of factors, ranging, inter alia, from environmental conditions, natural events, and demographics to both regional and global geopolitical contexts, economic conditions, energy sources, policies, and regulations. Given these, this paper examines the integration of news information from the global scene into Greek energy load forecasting schemes through the use of sentiment analysis. Investigating the ways the general emotional footprint of news worldwide affects and can be used in an energy modeling context, we benchmark possible ensemble configurations incorporating a multitude of 31 emotion polarities. Building on our previous work, an ensemble method that exploits specific outputs from a multi-label sentiment classifier and a sentiment analysis procedure under a multivariate forecasting scheme is presented. It is shown, through an empirical evaluation of the results, that the integration of emotion representations related to non-Greek news concerning global current affairs improves the predictions.

Speech assistive system for people with neuro disorders is a highly challenging task till date. Any kind of neuro cognitive disability affects the speech production mechanism that leads to speech impairment. Representation learning methods have recently emerged to improve the outcome of machine learning algorithms. In case of complex recognition tasks such as disordered speech recognition, learning compact and efficient representations for disordered speech utterances is important. Recent deep learning-based architectures need sufficiently large amount of impaired speech samples which are tedious with respect to neurologically disabled people. In this work, we focus on proposing a representation learning approach that uses traditional sequential model such as Hidden Markov Model (HMM) which works moderately well with small amount of impaired speech data. We propose a novel sequence to vector-based HMM State Sequence (HMM-SS) approach which is very compact and has proved to be an efficient representation for disordered speech utterances. The efficiency of the proposed HMM-SS approach is assessed using four datasets, namely 50 words of TORGO, 100-common words dataset of the UA-SPEECH, 50 help-seeking words and 100-common words of Impaired speech corpus in Tamil corpus. The proposed approach outperforms the baseline HMM, DNN-HMM and a recent state-of-the-art approach for all four datasets. The discriminative ability and the compactness of the proposed representation proved effective for disordered speech recognition.

Domain adaptation is an important area of research, as it aims to remedy the effects of the domain shift due to differences in the distribution between the source domain used for training and the target domain where prediction takes place. However, methods for characterizing the domain shift across datasets are lacking. In this work, we propose a domain shift metric called SpOT, which stands for spherical optimal transport, by operating on the spherical manifold. We realize our approach with a spherical network, used to obtain features, and an orthogonal projection loss, used to impose orthogonality in the feature space. The resulting spherical features have better inter-class separation and lower intra-class variation compared to features in Euclidean space. This type of feature clustering makes each domain representation more compact and more suitable for further analysis. The domain shift between the datasets is calculated using the optimal transport on the spherical features, which has a sound theoretical basis. Our results are further supported by experiments that show the correlation of SpOT with a new gain of transfer measure across domain adaptation datasets.

A vast amount of textual information on the internet has amplified the importance of text summarization models. Abstractive summarization generates original words and sentences that may not exist in the source document to be summarized. Such abstractive models may suffer from shortcomings such as linguistic acceptability and hallucinations. Recall-Oriented Understudy for Gisting Evaluation (ROUGE) is a metric commonly used to evaluate abstractive summarization models. However, due to its n-gram-based approach, it ignores several critical linguistic aspects. In this work, we propose Similarity, Entailment, and Acceptability Score (SEAScore), an automatic evaluation metric for evaluating abstractive text summarization models using the power of state-of-the-art pre-trained language models. SEAScore comprises three language models (LMs) that extract meaningful linguistic features from candidate and reference summaries and a weighted sum aggregator that computes an evaluation score. Experimental results show that our LM-based SEAScore metric correlates better with human judgment than standard evaluation metrics such as ROUGE-N and BERTScore.

Phishing attacks are increasing every year, both in terms of number and technique. Using only human weaknesses, an attacker can easily obtain the victim’s credentials or access their network. The problem persists despite many approaches offered by researchers, due to its dynamic nature, in which new phishing tactics are created every time. We, therefore, need more robust and effective methods to detect phishing emails. In this paper, we aim to detect phishing emails using the body text of the email with the hybrid approach combining case-based reasoning (CBR) and a deep learning model. Our proposed model, called DL-CBR, consists of a Bidirectional Long Short-Term Memory (Bi-LSTM) + Temporal Convolutional Network (TCN) network with an attention mechanism followed by a CBR classifier. The deep learning model is used for email representation, where it is trained using the $N$-pair loss function. To demonstrate the performance of DL-CBR, evaluation metrics, such as precision, accuracy, recall, and F-measure, were used, where we obtained an accuracy of 98.28%. The results show that our model outperformed other CBRs that utilize classical text representations like TF-IDF and Bag-of-Words. Additionally, while our model’s performance is slightly below that of the state-of-the-art models, it offers several advantages inherent to CBR. For instance, it can learn from new cases and update their database accordingly.