Qaiser Abbas

I am a Senior Machine Learning Engineer at Octopus Digital, where I work on machine learning, computer vision, data science, and NLP application development to solve complex business problems. Before joining Octopus I was working at SDSol Technologies on Generative AI, NLP and speech recognition projects in healthcare domain.

Previously, I have worked at Bioinformatics Research Lab and Computer Vision & Machine Learning Research Lab as a Research Assistant and Research Intern respectively in various Medical Image Computing and Image Recognition Projects.

From March 2021 to September 2022, I was assosiated with Department of Computer Science @ UET as an AI Instructor. I taught AI/ML course and conducted labs. I also conducted research in Medical Image Analysis. Our research proposal related to Deep Learning and Agriculture won research funding of PKR 3.5 million from HEC's NRPU program.

Email  /  CV  /  Scholar  /  Research Gate  /  Github  /  LinkedIn  /  ORCiD

I am broadly interested in learning from limited data for visual recognition problems. My primary area of interests include transfer learning and self supervised learning with applications to Medical Image Computing, Agriculture, and Vision based Industrial Defects Detection. I believe that artifical vision and learning from limited data can help us in improved object recognition, detection and segmentation to solve real-world problems.

• Deep Learning
• Medical Image Computing
• Computer Vision
• Self-Supervised Learning
• Natural Language Processing
• Applications of AI/ML

• [Sep 2024] Manuscript related to AIoT and Adversarial Attacks on video anomaly detection has been submitted to Q1 journal.
• [Sep 2023] Network Intrusion Detection Paper's PDF is now available online.
• [August 2023] Our paper related to network security and ML, entitled "Optimization of Predictive Performance of Intrusion Detection System Using Hybrid Ensemble Model for Secure Systems" has been accepted for publication in "PeerJ Computer Science".
• [May 2023] Punjabi NER Paper's PDF is now available online.
• [April 2023] A paper entitled "Using Data Augmentation and Bidirectional Encoder Representations from Transformers for Improving Punjabi Named Entity Recognition" has been accepted for publication in The ACM Transactions on Asian and Low-Resource Language Information Processing (TALLIP)
• [Novemeber 2022] I have joined SDSol Technologies as a Machine Learning Engineer. I will work on deep learning and computer vision projects.
• [October 2022] Part of my MS thesis published in SN Computer Science. Title: Detection and Classification of Malignant Melanoma Using Deep Features of NASNet
• [March 2022] Our research proposal for "Tea Plant Disease Prediction using Remote Sensing and Machine Learning" got funding of PKR 3.5M.
• [October 2021] First paper from my MS thesis work published. Title: Acral melanoma detection using dermoscopic images and convolutional neural networks
• [November 2021] I joined WortelAI as a part time Software Engineer (Deep Learning and Computer Vision)
• [Mar 2021] I joined Department of Computer Science as a Teaching Fellow @ UET Lahore

We experimented with various machine learning models for recent intrusion detection datasets. NSL-KDD, CSE-CIC-IDS2018 and UNSW-NB-15 datasets were used in the study. We proposed a hybrid, cost and compute efficient ensemble model for development of secure IDS systems.

We worked on data generation for Punjabi (Shahmukhi Script). A multilingual BERT model was trained for NER task. We developed a large training corpora using a simple and novel PoW augmentation technique

We investigtaed a NASNet neural network architecture for Malignant melanoma detection using dermoscopic images and data augmentation.

We proposed a neural network architecture for classification of acral lentiginous melanoma using dermoscopic images.

Due to acral melanoma’s infrequent occurrences, limited data is available so its early diagnosis is hard. To overcome this problem, we applied data centric techniques to develop a large dataset to train a deep learning models. Our proposed convolutional neural network achieved an accuracy of 91% on test set.