My name is Khawar Islam. I am working as a Deep Learning Researcher at COREMAX, Seoul, South Korea, working on Cutting-edge AI Research for Abusive Content Classifiers. I received my MS degree from Sejong University, advised by Prof. Hyeonjoon Moon and in collaboration with Prof. Mang Ye. Before joining COREMAX, I was a Student Researcher at CVPR Lab, working on Deep Learning Research for Sewer Defect Classification, Image Compression, Face Recognition and Vision Transformer.

Personal Website: https://khawar-islam.github.io/

vision transformer, continual learning, scene graphs, reward, inverse reinforcement learning, backpropogation, shape bias, knowledge distillation, self-supervised learning, augmented memory, variational auto-encoder, associative memory, foveation, Blender, cause-effect, neuro-symbolic, recurrent neural networks, LSTM, Mturk, Prolific, Python, Matlab, Pytorch, Tensorflow, NVIDIA, cross-entropy, maximum likelihood, metric learning, representation learning, open-set, long-tailed, object detection, object classification, mixup, style transfer, data augmentation, curriculum learning, generalization, out-of-distribution, domain shifts, random forest, decision tree, Monte Carlo tree search, hard and soft attention, meta-learning, depth estimation, video tracking, predictive coding