Sitemap
A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.
Pages
Posts
Multilingual speech recognition initiative for African languages
Published:
KEY INSIGHT
The paper discusses the need to address low-resource language gaps in speech recognition for African languages and proposes methods to improve language representation using modern machine learning techniques. It introduces the DVoice initiative, a community-driven initiative to create datasets and models for various African languages, focusing on accessibility and collaboration. The paper also presents various approaches for developing multilingual speech recognition systems, including innovative model training techniques, focusing on data quality and collection, and discussing the potential social and economic benefits of ASR systems for communities in Africa, particularly in improving access to technology for populations with high illiteracy rates.
The Nature of NLP: Analyzing Contributions in NLP Papers
Published:
KEY INSIGHT
The study uses pre-trained language models, including general-purpose models like BERT and RoBERTa, fine-tuned using the NLPContributions dataset. Large Language Models like GPT-3.5-Turbo and GPT-4-Turbo are used, fine-tuned with Reinforcement Learning from Human Feedback. A binary relevance method is used for classification, avoiding overfitting. A random baseline is used for comparison. Domain-specific models like BiomedBERT and SciBERT are used to analyze contributions in NLP research, with the SciBERT model specifically applied to research paper abstracts.
The State of the Art of Natural Language Processing — A Systematic Automated Review of NLP Literature Using NLP Techniques
Published:
KEY INSIGHT
The paper The State of the Art of Natural Language Processing-A Systematic Automated Review of NLP Literature Using NLP Techniques provides a comprehensive overview of the rapidly growing field of Natural Language Processing (NLP) in artificial intelligence. The study aims to capture meta-level knowledge, analyze existing literature, and promote transparency and accessibility through full automation. It also serves as a guide for using basic NLP tools, beneficial for researchers entering the field.
portfolio
Portfolio item number 1
Short description of portfolio item number 1
Portfolio item number 2
Short description of portfolio item number 2
publications
Parameter Estimation of Software Reliability Growth Models: A Comparison Between Grey Wolf Optimizer and Improved Grey Wolf Optimizer.
Published in IEEE, 2021
ABSTRACT
In modern era, industries demand for innovative and reliable software solutions. To maintain the reliability level of softwares various software reliability growth models were proposed in last four decades. These models performance relies on parameter estimation approaches utilized to find the optimum values of their unknown model parameters. But, developing an approach that provides the perfect optimum parameter for software reliability growth models (SRGMs) has been the issue of concern within the research community over the decades. This paper adopted the Improved Grey Wolf Optimizer (IGWO) for parameter estimation and compares its accuracy with existing approach Grey Wolf Optimizer (GWO) in estimating the optimum parameters for software reliability growth models. GWO imitates the social leadership pyramid and the hunting methods adopted by grey wolves; IGWO was later proposed to resolve the deficiencies observed in GWO for improved performance. Seven real world failure datasets have been utilized to measure and evaluate the performance of the proposed approach against the existing approach. The results indicate that proposed approach (IGWO) outperform the existing one (GWO).
The First Corpus for Detecting Fake News in Hausa Language
Published in SPRINGER, 2022
ABSTRACT
The rapid spread of deceptive news especially in Africa has become a global issue in last decade. This triggers the attention of the research community to develop efficient and reliable classification approaches for fake news detection so as to prevent its spread in the community. It has been explored that fake news in regional languages spread with a faster pace as compare to English language in local regions. Hausa is a very common language in Nigeria and some West African countries. So, it opens the challenge to detect the fake news in Hausa language. This paper presents the first corpus for the detection of fake news in Hausa. A dataset has been formed by collecting the labeled real and fake news consists of 2600 articles. In order to classify the fake news in Hausa, six different classifiers have been utilized. The performance of these approaches is then evaluated over different metrics and compared to determine the best model on the Hausa language dataset. The experimental results indicate that support vector machine (SVM) outperformed the other classifiers used by achieving 85% accuracy while AdaBoost happens to emerge as the fair model with 70% accuracy.
talks
Machine Learning Algorithms: Supervised Learning Tip to Tail
Published:
The Machine Learning Algorithms: Supervised Learning Tip to Tail course on Coursera provides a deep dive into the foundations and practical applications of supervised learning. It is designed to help participants understand how supervised learning works, its real-world relevance, and how to implement and optimize these algorithms. The course begins by explaining the core concepts of supervised learning, such as labeled data, target prediction, and evaluation metrics. Learners are introduced to common tasks like classification and regression, which form the basis of predictive modeling in a wide range of industries.
Introduction to Data Science in Python
Published:
The Introduction to Data Science in Python course on Coursera offers a comprehensive entry point into the world of data science, focusing on Python as the primary tool for data analysis. The course begins by introducing participants to the fundamental concepts of data science, such as data wrangling, exploration, and visualization. Learners are guided through Python’s essential libraries, such as pandas for data manipulation and NumPy for numerical computations, building a solid foundation for handling and processing structured data effectively.
Natural Language Processing
Published:
The Natural Language Processing course on Coursera is a comprehensive introduction to the field of NLP, designed to equip participants with the knowledge and tools to work on text and language data. The course begins by explaining the fundamentals of NLP, including tokenization, stemming, lemmatization, and text preprocessing. Learners are introduced to the unique challenges of understanding and processing human language, such as ambiguity, context, and syntax, and how computational methods can address these complexities.
Transfer Learning for NLP with TensorFlow Hub
Published:
The Transfer Learning for NLP with TensorFlow Hub course on Coursera introduces participants to the powerful concept of transfer learning in natural language processing (NLP). Transfer learning allows developers to leverage pre-trained models, reducing the need for massive datasets and extensive computational resources. Participants learn why this technique is particularly effective for NLP, where tasks like sentiment analysis, text classification, or entity recognition require understanding complex linguistic patterns. The course emphasizes how pre-trained models simplify these challenges by encoding linguistic knowledge learned from large-scale datasets.
Fine Tune BERT for Text Classification with TensorFlow
Published:
Fine Tune BERT for Text Classification with TensorFlow offers a hands-on introduction to the practical application of transfer learning using BERT (Bidirectional Encoder Representations from Transformers) for natural language processing tasks, specifically text classification. Learners are introduced to the fundamentals of BERT, including how it works, why it is effective for NLP tasks, and its advantages over traditional machine learning approaches for textual data. By the end of the course, students will have a clear understanding of how pretrained models can be fine-tuned for specific tasks, saving time and resources compared to training models from scratch.
Tweet Emotion Recognition with TensorFlow
Published:
Tweet Emotion Recognition with TensorFlow provides an applied introduction to building deep learning models for understanding emotions expressed in tweets. It focuses on leveraging TensorFlow and Natural Language Processing (NLP) techniques to detect emotions such as happiness, sadness, anger, and more from social media text. Learners explore how emotion recognition can enhance sentiment analysis, improve user engagement, and support applications like mental health monitoring or customer feedback analysis.
teaching
TEACHING EXPERIENCE
, Aminu Kano College of Islamic and Legal Studies Kano, Department of Computer Science, 2014
I am a dedicated computer scientist with an M.Sc and B.Sc in Computer Science, currently pursuing a Ph.D. in the field. My expertise spans natural language processing, machine learning, and deep learning, with a special focus on low-resource languages. I am also an experienced educator, teaching C++, Java, and Python, and I have a passion for advancing AI research and writing journal articles.