Ananjan Maiti
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Ananjan Maiti has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
Biswajoy Chatterjee
University of Engineering and Management (UEM)
3 shared publications- Mahamuda Sultana
Guru Nanak Institute of Technology
2 shared publications - Suman Bhattacharya
Guru Nanak Institute of Technology
2 shared publications Amira S. Ashour
Department of Electronics and Electrical Communications Engineering, Faculty of Engineering, Tanta University, Tanta, Egypt
1 shared publicationAmit Kumar Bhore
Department of Computer Applications, Guru Nanak Institute of Technology, Kolkata, India
1 shared publicationArpita Dutta
Techno India University
1 shared publication
Latest Publications
- 15 May 2026
- Published by Bentham Science Publishers Ltd. in Current Computer-Aided Drug Design
- Vol. 22, 1-14
Augmented Chemical Language Meets Descriptor Space: A Hybrid Deep-learning Pipeline for Predicting Blood-brain Barrier Penetration of Drug-like Molecules
Introduction/Objective: The brain has a semipermeable interface, the blood-brain barrier (BBB), which protects the brain but greatly limits the absorption of most small-molecule drugs, with only ~ 24% able to cross. Methods: This study involved 5,412 drug-like molecules from three public BBB datasets and calculated 200 physicochemical descriptors with RDKit and Mordred. With ChemBERTa-2, SMILES strings were tokenized using the ChemBERTa-2 tokenizer and embedded into a 768-dimensional chemical language space. A label-aware SMILES mask-and-replace augmentation expanded the corpus fivefold while retaining chemical validity. Using Bemis-Murcko scaffolds and 5-fold cross-validation on a 4-layer MLP, descriptor and embedding vectors were combined into a 968-dimensional hybrid representation. Statistical reliability was assessed using 2,000-sample bootstrap confidence intervals. Results: The hybrid model (accuracy, 95%) and an AUROC of 0.96 showed an improvement of 7-11% above the descriptors-only (accuracy 0.88; AUROC 0.90) or embedding-only (accuracy 0.91; AUROC 0.93) baselines. Statistically significant (p < 0.001) gains were supported by saliency analyses of descriptors and embeddings, which made unique and orthogonal contributions. Discussion: The findings indicate that integrating traditional descriptors and chemical-language embeddings yields two distinct types of information, enabling strong, understandable predictions. Augmentation strategies increased performance more quickly without causing syntactic invalidity. Conclusion: A hybrid AI pipeline shows much better BBB penetration forecasting than single- baseline unimodal baselines. This work facilitates reproducibility by releasing cleaned data, trained weights, and open-source code, and by providing a viable resource for ongoing CNS drug discovery.
- 15 May 2026
- Published by Bentham Science Publishers Ltd. in Current Computer-Aided Drug Design
- Vol. 22, 1-14
Augmented Chemical Language Meets Descriptor Space: A Hybrid Deep-learning Pipeline for Predicting Blood-brain Barrier Penetration of Drug-like Molecules
Introduction/Objective: The brain has a semipermeable interface, the blood-brain barrier (BBB), which protects the brain but greatly limits the absorption of most small-molecule drugs, with only ~ 24% able to cross. Methods: This study involved 5,412 drug-like molecules from three public BBB datasets and calculated 200 physicochemical descriptors with RDKit and Mordred. With ChemBERTa-2, SMILES strings were tokenized using the ChemBERTa-2 tokenizer and embedded into a 768-dimensional chemical language space. A label-aware SMILES mask-and-replace augmentation expanded the corpus fivefold while retaining chemical validity. Using Bemis-Murcko scaffolds and 5-fold cross-validation on a 4-layer MLP, descriptor and embedding vectors were combined into a 968-dimensional hybrid representation. Statistical reliability was assessed using 2,000-sample bootstrap confidence intervals. Results: The hybrid model (accuracy, 95%) and an AUROC of 0.96 showed an improvement of 7-11% above the descriptors-only (accuracy 0.88; AUROC 0.90) or embedding-only (accuracy 0.91; AUROC 0.93) baselines. Statistically significant (p < 0.001) gains were supported by saliency analyses of descriptors and embeddings, which made unique and orthogonal contributions. Discussion: The findings indicate that integrating traditional descriptors and chemical-language embeddings yields two distinct types of information, enabling strong, understandable predictions. Augmentation strategies increased performance more quickly without causing syntactic invalidity. Conclusion: A hybrid AI pipeline shows much better BBB penetration forecasting than single- baseline unimodal baselines. This work facilitates reproducibility by releasing cleaned data, trained weights, and open-source code, and by providing a viable resource for ongoing CNS drug discovery.
- 24 November 2025
- Published by Springer Nature
- p. 314-324
Unsupervised Pattern Discovery in Cyber Incidents Using Principal Component Analysis K-Means DBSCAN and Isolation Forest
- 24 November 2025
- Published by Springer Nature
- p. 314-324
Unsupervised Pattern Discovery in Cyber Incidents Using Principal Component Analysis K-Means DBSCAN and Isolation Forest
- 29 July 2025
- Published by Springer Nature
- p. 433-445
The Impact of AI-Enhanced Chatbots on Psychological Support for Higher Education Students
- 29 July 2025
- Published by Springer Nature
- p. 433-445
The Impact of AI-Enhanced Chatbots on Psychological Support for Higher Education Students
- 21 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Identification of Bird via Acoustics for Bird Species Population Monitoring and Conservation
Nowadays, bird conservation has become a crucial part of the study. The paramount peril confronting avian species in jeopardy pertains to the degradation of their habitats. The chirping sound of birds can be heard with lots of noise and environmental sounds. So, monitoring and identification of bird sounds are more challenging. This study proposes meticulous monitoring and identifying bird sounds to ensure bird species' conservation and automate the remote monitoring of bird populations. Firstly, the data frame was generated from the acquired recorded dataset. Then, the data frames were segmented into slices of 0.5-second intervals so that the silent zones of the recorded dataset could be removed. Following this, unwanted noise was removed using the filter. Then, the Mel Frequency Cepstral Coefficients (MFCCs) spectrogram was generated using feature extraction methodology. From this spectrogram, a sound vector is generated. Monitoring and tuning different hyperparameter constraints using Principal Component Analysis (PCA), the model was subsequently trained using Dense Neural Network. The test demonstrates exceptional precision and reliability with a nominal margin of error.
- 21 November 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Identification of Bird via Acoustics for Bird Species Population Monitoring and Conservation
Nowadays, bird conservation has become a crucial part of the study. The paramount peril confronting avian species in jeopardy pertains to the degradation of their habitats. The chirping sound of birds can be heard with lots of noise and environmental sounds. So, monitoring and identification of bird sounds are more challenging. This study proposes meticulous monitoring and identifying bird sounds to ensure bird species' conservation and automate the remote monitoring of bird populations. Firstly, the data frame was generated from the acquired recorded dataset. Then, the data frames were segmented into slices of 0.5-second intervals so that the silent zones of the recorded dataset could be removed. Following this, unwanted noise was removed using the filter. Then, the Mel Frequency Cepstral Coefficients (MFCCs) spectrogram was generated using feature extraction methodology. From this spectrogram, a sound vector is generated. Monitoring and tuning different hyperparameter constraints using Principal Component Analysis (PCA), the model was subsequently trained using Dense Neural Network. The test demonstrates exceptional precision and reliability with a nominal margin of error.
- Open Access
- 1 January 2023
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 2876 (1)
Detection of skin cancer through shape features and grid search machine learning approach
Melanoma is the most common prevalent and lethal form of skin cancer. It could be curable, if it was discovered earlier. The discovery of melanocytes in the locality of the epidermis is a vital step in the diagnosis of melanoma. In our study, we utilize a mix of artificial intelligence learning to help dermatologists in the early diagnosis of melanoma. This study focuses on the active contour segmentation technique for separating objects. Extracted segmented parts undergo feature extraction for form feature identification. Twelve shape-based properties were retrieved as features. The acquired characteristics are then utilized to categorize the data. It was determined that the Random Forest grid search strategy was the best suited classifier. Among various classifiers, variables offered the best results for the categorization of skin lesions. With the application of hyper parameter modification, the suggested technique attained 92.8 percent accuracy. This study will be developed in the future utilizing a variety of machine learning algorithms to select the design for the bigger dataset. In the future, there could be more approaches that might be applied in this sort of investigation.
- 1 January 2023
- Published by AIP Publishing in AIP Conference Proceedings
- Vol. 2876 (1)
Open AccessDetection of skin cancer through shape features and grid search machine learning approach
Melanoma is the most common prevalent and lethal form of skin cancer. It could be curable, if it was discovered earlier. The discovery of melanocytes in the locality of the epidermis is a vital step in the diagnosis of melanoma. In our study, we utilize a mix of artificial intelligence learning to help dermatologists in the early diagnosis of melanoma. This study focuses on the active contour segmentation technique for separating objects. Extracted segmented parts undergo feature extraction for form feature identification. Twelve shape-based properties were retrieved as features. The acquired characteristics are then utilized to categorize the data. It was determined that the Random Forest grid search strategy was the best suited classifier. Among various classifiers, variables offered the best results for the categorization of skin lesions. With the application of hyper parameter modification, the suggested technique attained 92.8 percent accuracy. This study will be developed in the future utilizing a variety of machine learning algorithms to select the design for the bigger dataset. In the future, there could be more approaches that might be applied in this sort of investigation.
- 16 November 2022
- Published by Springer Nature in Innovations in Systems and Software Engineering
- Vol. 21 (1) , 151-164
Detection of skin cancer through hybrid color features and soft voting ensemble classifier
- 16 November 2022
- Published by Springer Nature in Innovations in Systems and Software Engineering
- Vol. 21 (1) , 151-164
Detection of skin cancer through hybrid color features and soft voting ensemble classifier
- 2 April 2021
- Published by Springer Nature
- p. 123-133
The Effect of Different Feature Selection Methods for Classification of Melanoma
- 2 April 2021
- Published by Springer Nature
- p. 123-133
The Effect of Different Feature Selection Methods for Classification of Melanoma
- 9 September 2020
- Published by Bentham Science Publishers Ltd. in Current Medical Imaging
- Vol. 16 (7) , 835-854
Computer-aided Diagnosis of Melanoma: A Review of Existing Knowledge and Strategies
Computer-aided diagnosis (CAD) systems are the best alternative for immediate disclosure and diagnosis of skin diseases. Such systems comprise several image processing procedures, including segmentation, feature extraction and artificial intelligence (AI) based methods. This survey highlights different CAD methodologies for diagnosing Melanoma and related skin diseases. It has also discussed types, stages, treatments and various imaging techniques of skin cancer. Currently, researchers developed new techniques to detect each stage. Extensive studies on melanoma cancer detection were performed by incorporating advanced machine learning. Still, there is a high need for an accurate, faster, affordable, portable methodology for a CAD system. This will strengthen the work in related fields and address the future direction of a similar kind of research.
- 9 September 2020
- Published by Bentham Science Publishers Ltd. in Current Medical Imaging
- Vol. 16 (7) , 835-854
Computer-aided Diagnosis of Melanoma: A Review of Existing Knowledge and Strategies
Computer-aided diagnosis (CAD) systems are the best alternative for immediate disclosure and diagnosis of skin diseases. Such systems comprise several image processing procedures, including segmentation, feature extraction and artificial intelligence (AI) based methods. This survey highlights different CAD methodologies for diagnosing Melanoma and related skin diseases. It has also discussed types, stages, treatments and various imaging techniques of skin cancer. Currently, researchers developed new techniques to detect each stage. Extensive studies on melanoma cancer detection were performed by incorporating advanced machine learning. Still, there is a high need for an accurate, faster, affordable, portable methodology for a CAD system. This will strengthen the work in related fields and address the future direction of a similar kind of research.
- 10 June 2019
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 79 (21-22) , 15635-15654
Improving detection of Melanoma and Naevus with deep neural networks
- 10 June 2019
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 79 (21-22) , 15635-15654
Improving detection of Melanoma and Naevus with deep neural networks