Dr. Abul Kalam Al Azad (original) (raw)

Papers by Dr. Abul Kalam Al Azad

International Journal of Electrical and Computer Engineering (IJECE)

The growth of abnormal cells in the brain’s tissue causes brain tumors. Brain tumors are consider... more The growth of abnormal cells in the brain’s tissue causes brain tumors. Brain tumors are considered one of the most dangerous disorders in children and adults. It develops quickly, and the patient’s survival prospects are slim if not appropriately treated. Proper treatment planning and precise diagnoses are essential to improving a patient’s life expectancy. Brain tumors are mainly diagnosed using magnetic resonance imaging (MRI). As part of a convolution neural network (CNN)-based illustration, an architecture containing five convolution layers, five max-pooling layers, a Flatten layer, and two dense layers has been proposed for detecting brain tumors from MRI images. The proposed model includes an automatic feature extractor, modified hidden layer architecture, and activation function. Several test cases were performed, and the proposed model achieved 98.6% accuracy and 97.8% precision score with a low cross-entropy rate. Compared with other approaches such as adjacent feature pro...

We represented a new Bangla dataset with a Hybrid Recurrent Neural Network model which generated ... more We represented a new Bangla dataset with a Hybrid Recurrent Neural Network model which generated Bangla natural language description of images. This dataset achieved by a large number of images with classification and containing natural language process of images. We conducted experiments on our self-made Bangla Natural Language Image to Text (BNLIT) dataset. Our dataset contained 8,743 images. We made this dataset using Bangladesh perspective images. We used one annotation for each image. In our repository, we added two types of pre-processed data which is 224 × 224 and 500 × 375 respectively alongside annotations of full dataset. We also added CNN features file of whole dataset in our repository which is features.pkl. Online Link - 1. http://doi.org/10.5281/zenodo.3372752 (Zenodo); 2. https://doi.org/10.7910/DVN/DZZ1ZB (Harvard Dataverse); 3. http://dx.doi.org/10.17632/ws3r82gnm8.1 (Mendeley); 4. https://www.kaggle.com/jishan900/bangla-natural-language-image-to-text-bnlit (Kaggle)

Indonesian Journal of Electrical Engineering and Computer Science, 2021

Automatic image captioning task in different language is a challenging task which has not been we... more Automatic image captioning task in different language is a challenging task which has not been well investigated yet due to the lack of dataset and effective models. It also requires good understanding of scene and contextual embedding for robust semantic interpretation of images for natural language image descriptor. To generate image descriptor in Bangla, we created a new Bangla dataset of images paired with target language label, named as Bangla Natural Language Image to Text (BNLIT) dataset. To deal with the image understanding, we propose a hybrid encoder-decoder model based on encoder-decoder architecture and the model is evaluated on our newly created dataset. This proposed approach achieves significance performance improvement on task of semantic retrieval of images. Our hybrid model uses the Convolutional NeuralNetwork as an encoder whereas the Bidirectional Long Short Term Memory is used for the sentence representation that decreases the computational complexities without ...

As an on-going pandemic caused by the out-break of severe acute respiratory syndrome coronavirus ... more As an on-going pandemic caused by the out-break of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) or simply COVID-19 sweeps through the globe at an unprecedented rate leaving behind trails of high infection and mortality, it is crucial to understand the propagation dynamics of the virus in a host population in order to take urgent and effective remedial and mitigating steps to save life. It is already observed in many countries and communities that accurate and timely testing, tracing, and tracking of the infection lead to better containment and slowing down of the spread. In this exploratory research, the early growth dynamics of infection within a population is pursued based on real data. The study posits that the early growth in a homogenous population follows an exponential pattern motivating further rigorous quantitative treatment based on a number of analytical models such as logistic model, Richard’s model, and Gompertz model– the acceleration pattern of the out...

International Journal of Innovative Technology and Exploring Engineering

The existing real time object detection algorithm is based on the deep neural network of convolut... more The existing real time object detection algorithm is based on the deep neural network of convolution need to perform multilevel convolution and pooling operations on the entire image to extract a deep semantic characteristic of the image. The detection models perform better for large objects. However, these models do not detect small objects with low resolution and noise, because the features of existing models do not fully represent the essential features of small objects after repeated convolution operations. We have introduced a novel real time detection algorithm which employs upsampling and skip connection to extract multiscale features at different convolution levels in a learning task resulting a remarkable performance in detecting small objects. The detection precision of the model is shown to be higher and faster than that of the state-of-the-art models.

2016 3rd International Conference on Electrical Engineering and Information Communication Technology (ICEEICT), 2016

2016 International Workshop on Computational Intelligence (IWCI), 2016

Frontiers in Behavioral Neuroscience, 2012

Frontiers in Neuroinformatics, 2011

In this paper we develop a computational model of the anatomy of a spinal cord. We address a long... more In this paper we develop a computational model of the anatomy of a spinal cord. We address a long-standing ambition of neuroscience to understand the structure-function problem by modeling the complete spinal cord connectome map in the 2-day old hatchling Xenopus tadpole. Our approach to modeling neuronal connectivity is based on developmental processes of axon growth. A simple mathematical model of axon growth allows us to reconstruct a biologically realistic connectome of the tadpole spinal cord based on neurobiological data. In our model we distribute neuron cell bodies and dendrites on both sides of the body based on experimental measurements. If growing axons cross the dendrite of another neuron, they make a synaptic contact with a defined probability. The total neuronal network contains ∼1,500 neurons of six cell-types with a total of ∼120,000 connections. The anatomical model contains random components so each repetition of the connectome reconstruction procedure generates a different neuronal network, though all share consistent features such as distributions of cell bodies, dendrites, and axon lengths. Our study reveals a complex structure for the connectome with many interesting specific features including contrasting distributions of connection length distributions. The connectome also shows some similarities to connectivity graphs for other animals such as the global neuronal network of C. elegans. In addition to the interesting intrinsic properties of the connectome, we expect the ability to grow and analyze a biologically realistic spinal cord connectome will provide valuable insights into the properties of the real neuronal networks underlying simple behavior.

Journal of Neuroscience, 2014

How do the pioneer networks in the axial core of the vertebrate nervous system first develop? Fun... more How do the pioneer networks in the axial core of the vertebrate nervous system first develop? Fundamental to understanding any full-scale neuronal network is knowledge of the constituent neurons, their properties, synaptic interconnections, and normal activity. Our novel strategy uses basic developmental rules to generate model networks that retain individual neuron and synapse resolution and are capable of reproducing correct, whole animal responses. We apply our developmental strategy to young Xenopus tadpoles, whose brainstem and spinal cord share a core vertebrate plan, but at a tractable complexity. Following detailed anatomical and physiological measurements to complete a descriptive library of each type of spinal neuron, we build models of their axon growth controlled by simple chemical gradients and physical barriers. By adding dendrites and allowing probabilistic formation of synaptic connections, we reconstruct network connectivity among up to 2000 neurons. When the resulting "network" is populated by model neurons and synapses, with properties based on physiology, it can respond to sensory stimulation by mimicking tadpole swimming behavior. This functioning model represents the most complete reconstruction of a vertebrate neuronal network that can reproduce the complex, rhythmic behavior of a whole animal. The findings validate our novel developmental strategy for generating realistic networks with individual neuron-and synapse-level resolution. We use it to demonstrate how early functional neuronal connectivity and behavior may in life result from simple developmental "rules," which lay out a scaffold for the vertebrate CNS without specific neuron-to-neuron recognition.

SIAM Journal on Applied Dynamical Systems, 2010

We study the appearance of a novel phenomenon for linearly coupled identical bursters: synchroniz... more We study the appearance of a novel phenomenon for linearly coupled identical bursters: synchronized bursts where there are changes of spike synchrony within each burst. The examples we study are for normal form elliptic bursters where there is a periodic slow passage through a Bautin (codimension two degenerate Andronov-Hopf) bifurcation. This burster has a subcritical Andronov-Hopf bifurcation at the onset of repetitive spiking while end of burst occurs via a fold limit cycle bifurcation. We study synchronization behavior of two and three Bautin-type elliptic bursters for a linear direct coupling scheme. Burst synchronization is known to be prevalent behavior among such coupled bursters, while spike synchronization is more dependent on the details of the coupling. We note that higher order terms in the normal form that do not affect the behavior of a single burster can be responsible for changes in synchrony pattern; more precisely, we find within-burst synchrony changes associated with a turning point in the spiking frequency.

PLoS ONE, 2014

Relating structure and function of neuronal circuits is a challenging problem. It requires demons... more Relating structure and function of neuronal circuits is a challenging problem. It requires demonstrating how dynamical patterns of spiking activity lead to functions like cognitive behaviour and identifying the neurons and connections that lead to appropriate activity of a circuit. We apply a ''developmental approach'' to define the connectome of a simple nervous system, where connections between neurons are not prescribed but appear as a result of neuron growth. A gradient based mathematical model of two-dimensional axon growth from rows of undifferentiated neurons is derived for the different types of neurons in the brainstem and spinal cord of young tadpoles of the frog Xenopus. Model parameters define a twodimensional CNS growth environment with three gradient cues and the specific responsiveness of the axons of each neuron type to these cues. The model is described by a nonlinear system of three difference equations; it includes a random variable, and takes specific neuron characteristics into account. Anatomical measurements are first used to position cell bodies in rows and define axon origins. Then a generalization procedure allows information on the axons of individual neurons from small anatomical datasets to be used to generate larger artificial datasets. To specify parameters in the axon growth model we use a stochastic optimization procedure, derive a cost function and find the optimal parameters for each type of neuron. Our biologically realistic model of axon growth starts from axon outgrowth from the cell body and generates multiple axons for each different neuron type with statistical properties matching those of real axons. We illustrate how the axon growth model works for neurons with axons which grow to the same and the opposite side of the CNS. We then show how, by adding a simple specification for dendrite morphology, our model ''developmental approach'' allows us to generate biologically-realistic connectomes.

International Journal of Advances in Intelligent Informatics, Jul 12, 2020

Automated image to text generation is a computationally challenging computer vision task which re... more Automated image to text generation is a computationally challenging computer vision task which requires sufficient comprehension of both syntactic and semantic meaning of an image to generate a meaningful description. Until recent times, it has been studied to a limited scope due to the lack of visual-descriptor dataset and functional models to capture intrinsic complexities involving features of an image. In this study, a novel dataset was constructed by generating Bangla textual descriptor from visual input, called Bangla Natural Language Image to Text (BNLIT), incorporating 100 classes with annotation. A deep neural network-based image captioning model was proposed to generate image description. The model employs Convolutional Neural Network (CNN) to classify the whole dataset, while Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) capture the sequential semantic representation of text-based sentences and generate pertinent description based on the modular complexities of an image. When tested on the new dataset, the model accomplishes significant enhancement of centrality execution for image semantic recovery assignment. For the experiment of that task, we implemented a hybrid image captioning model, which achieved a remarkable result for a new self-made dataset, and that task was new for the Bangladesh perspective. In brief, the model provided benchmark precision in the characteristic Bangla syntax reconstruction and comprehensive numerical analysis of the model execution results on the dataset.

International Journal of Electrical and Computer Engineering (IJECE), 2019

We presented a learning model that generated natural language description of images. The model ut... more We presented a learning model that generated natural language description of images. The model utilized the connections between natural language and visual data by produced text line based contents from a given image. Our Hybrid Recurrent Neural Network model is based on the intricacies of Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Bi-directional Recurrent Neural Network (BRNN) models. We conducted experiments on three benchmark datasets, e.g., Flickr8K, Flickr30K, and MS COCO. Our hybrid model utilized LSTM model to encode text line or sentences independent of the object location and BRNN for word representation, this reduced the computational complexities without compromising the accuracy of the descriptor. The model produced better accuracy in retrieving natural language based description on the dataset.

Institute of Advanced Engineering and Science (IAES), Aug 1, 2019