Mandar Gogate

Persian Named Entity Recognition

Named Entity Recognition (NER) is an important natural language processing (NLP) tool for information extraction and retrieval from unstructured texts such as newspapers, blogs and emails. NER involves processing unstructured text for classification of words or expressions into relevant categories. In literature, NER has been developed for various languages but limited work has been conducted to develop NER for Persian text. This is due to limited resources (such as corpus, lexicons etc.) and tools for Persian named entities. In this paper, a novel scalable system for Persian Named Entity Recognition (PNER) is presented. The proposed PNER can recognize and extract three most important named entities in Persian script: the person name, location and date. The proposed PNER has been developed by combining a grammatical rule-based approach with machine learning. The proposed framework has integrated dictionaries of Persian named entities, Persian grammar rules and a Support Vector Machine (SVM). The performance evaluation of PNER in terms of precision, recall and f-measure has achieved comparable results with the state-of-the-art NER frameworks in other languages.

A Comparative Study of Persian Sentiment Analysis Based on Different Feature Combinations

In recent years, the use of internet and correspondingly the number of online reviews, comments and opinions have increased significantly. It is indeed very difficult for humans to read these opinions and classify them accurately. Consequently, there is a need for an automated system to process this big data. In this paper, a novel sentiment analysis framework for Persian language has been proposed. The proposed framework comprises three basic steps: pre-processing, feature extraction, and support vector machine (SVM) based classification. The performance of the proposed framework has been evaluated taking into account different features combinations. The simulation results have revealed that the best performance could be achieved by integrating unigram, bigram, and trigram features.

A Survey on the Role of Wireless Sensor Networks and IoT in Disaster Management

Extreme events and disasters resulting from climate change or other ecological factors are difficult to predict and manage. Current limitations of state-of-the-art approaches to disaster prediction and management could be addressed by adopting new unorthodox risk assessment and management strategies. The next generation Internet of Things (IoT), Wireless Sensor Networks (WSNs), 5G wireless communication, and big data analytics technologies are the key enablers for future effective disaster management infrastructures. In this chapter, we commissioned a survey on emerging wireless communication technologies with potential for enhancing disaster prediction, monitoring, and management systems. Challenges, opportunities, and future research trends are highlighted to provide some insight on the potential future work for researchers in this field.

Toward’s Arabic Multi-modal Sentiment Analysis

In everyday life, people use internet to express and share opinions, facts, and sentiments about products and services. In addition, social media applications such as Facebook, Twitter, WhatsApp, Snapchat etc., have become important information sharing platforms. Apart from these, a collection of product reviews, facts, poll information, etc., is a need for every company or organization ranging from start-ups to big firms and governments. Clearly, it is very challenging to analyse such big data to improve products, services, and satisfy customer requirements. Therefore, it is necessary to automate the evaluation process using advanced sentiment analysis techniques. Most of previous works focused on uni-modal sentiment analysis mainly textual model. In this paper, a novel Arabic multimodal dataset is presented and validated using state-of-the-art support vector machine (SVM) based classification method.

A novel brain-inspired compression-based optimised multimodal fusion for emotion recognition

The curse of dimensionality is a well-established phenomenon. However, the properties of high dimensional data are often poorly understood and overlooked during the process of data modelling and analysis. Similarly, how to optimally fuse different modalities is still a big research question. In this paper, we addressed these challenges by proposing a novel two level brain-inspired compression based optimised multimodal fusion framework for emotion recognition. In the first level, the framework extracts the compressed and optimised multimodal features by applying a deep convolutional neural network (CNN) based compression on each modality (i.e. audio, text, and visuals). The second level simply concatenates the extracted optimised and compressed features for classification. The performance of the proposed approach with two different compression levels (i.e. 78% and 98%) is compared with late fusion (class level-1 dimension, class probabilities level-4 dimension) and early fusion (feature level-72000 dimension). The simulation results and critical analysis have demonstrated up to 10% and 5% performance improvement as compared to the state-of-the-art support vector machine (SVM) and long-short-term memory (LSTM) based multimodal emotion recognition systems respectively. We hypothesise that there exist an optimal level of compression at which optimised multimodal features could be extracted from each modality, which could lead to a significant performance improvement.

Complex-valued computational model of hippocampal CA3 recurrent collaterals

Complex planes are known to simplify the complexity of real world problems, providing a better comprehension of their functionality and design. The need for complex numbers in both artificial and biological neural networks is equally well established. In the latter, complex numbers allows neuroscientists to consider and analyze the phase component of brain oscillations occurring during chains of action potentials. This paper implements complex-valued weights and inputs in the real valued recurrent collaterals model introduced by Káli & Dayan for the CA3 region of the hippocampus, with equations appropriately modified to include the phase component. Complex models can generally be implemented by solving the real and complex parts separately resulting from solving the model equations twice. This implementation is simulated here and the results demonstrate the models potential utility for further mathematical and neurobiological analysis to define a proper phase function which oscillates in the theta frequency range.

Exploiting Deep Learning for Persian Sentiment Analysis

The rise of social media is enabling people to freely express their opinions about products and services. The aim of sentiment analysis is to automatically determine subject’s sentiment (e.g., positive, negative, or neutral) towards a particular aspect such as topic, product, movie, news etc. Deep learning has recently emerged as a powerful machine learning technique to tackle a growing demand of accurate sentiment analysis. However, limited work has been conducted to apply deep learning algorithms to languages other than English, such as Persian. In this work, two deep learning models (deep autoencoders and deep convolutional neural networks (CNNs)) are developed and applied to a novel Persian movie reviews dataset. The proposed deep learning models are analyzed and compared with the state-of-the-art shallow multilayer perceptron (MLP) based machine learning model. Simulation results demonstrate the enhanced performance of deep learning over state-of-the-art MLP.

Statistical Analysis Driven Optimized Deep Learning System for Intrusion Detection

Attackers have developed ever more sophisticated and intelligent ways to hack information and communication technology (ICT) systems. The extent of damage an individual hacker can carry out upon infiltrating a system is well understood. A potentially catastrophic scenario can be envisaged where a nation-state intercepting encrypted financial data gets hacked. Thus, intelligent cybersecurity systems have become inevitably important for improved protection against malicious threats. However, as malware attacks continue to dramatically increase in volume and complexity, it has become ever more challenging for traditional analytic tools to detect and mitigate threat. Furthermore, a huge amount of data produced by large networks have made the recognition task even more complicated and challenging. In this work, we propose an innovative statistical analysis driven optimized deep learning system for intrusion detection. The proposed intrusion detection system (IDS) extracts optimized and more correlated features using big data visualization and statistical analysis methods, followed by a deep autoencoder (AE) for potential threat detection. Specifically, a preprocessing module eliminates the outliers and converts categorical variables into one-hot-encoded vectors. The feature extraction module discards features with null values grater than 80% and selects the most significant features as input to the deep autoencoder model trained in a greedy-wise manner. The NSL-KDD dataset (an improved version of the original KDD dataset) from the Canadian Institute for Cybersecurity is used as a benchmark to evaluate the feasibility and effectiveness of the proposed architecture. Simulation results demonstrate the potential of our proposed IDS system for improving intrusion detection as compared to existing state-of-the-art methods.