Muhammad Kamran Janjua

Art is Long; Time is Fleeting!

I am a graduate student with Dr. Martha White at University of Alberta, where I work on representation learning for reinforcement learning. Particularly I work at the combination of offline and online RL where the goal is to learn effective representations in the offline phase to warm-start the agent in the online setting and make sure that the agent converges quickly.

Previously, I worked as a Research Associate at Qatar Computing Research Institute (QCRI) working on interpretability in neural language models. Specifically, I worked on understanding how neural language models build (if at all) grammatical structure of language internally. I was supervised by Dr. Hassan Sajjad.

Before that I did my undergraduate from National University of Sciences & Technology (NUST), Islamabad working in TUKL-NUST R&D Lab working with Dr. Faisal Shafait. Over the years in this pursuit of science, I have had the pleasure of visiting AaltoVision and AaltoML lab at Aalto University, Finland to work on estimating depth from stereo setups. I have also worked as a research intern at ARTE Lab, University of Insubria back in 2018 where I worked on designing multi-modal neural networks. My research was supervised by Dr. Ignazio Gallo.

Email  /  CV  /  Quora  /  Google Scholar  /  Github

profile photo

Although I have worked mainly with computer vision, I am now interested in reinforcement learning with a focus on representation learning techniques. I work on developing methods to effectively learn representations that can be distilled in upstream tasks.

b3do Movement-induced Priors for Deep Stereo
Yuxin Hou, Muhammad Kamran Janjua, Juho Kannala, Arno Solin
25th International Conference on Pattern Recognition (ICPR), 2020

We propose a method for fusing stereo disparity estimation with movement-induced prior information.

b3do Do Cross Modal Systems Leverage Semantic Relationships?
Shah Nawaz, Muhammad Kamran Janjua*, Ignazio Gallo, Arif Mahmood, Alessandro Calefati, Faisal Shafait
Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2019

We propose a new measure SemanticMap to evaluate the performance of cross modal systems. Our proposed measure evaluates the semantic similarity between the image and text representations in the latent embedding space.

b3do Deep latent space learning for cross-modal mapping of audio and visual signals
Shah Nawaz*, Muhammad Kamran Janjua*, Ignazio Gallo, Arif Mahmood, Alessandro Calefati
Digital Image Computing: Techniques and Applications (DICTA), 2019

We propose a novel deep training algorithm for joint representation of audio and visual information which consists of a single stream network (SSNet) coupled with a novel loss function to learn a shared deep latent space representation of multimodal information.

b3do Git Loss for Deep Face Recognition
Alessandro Calefati*, Muhammad Kamran Janjua*, Shah Nawaz, Ignazio Gallo
British Machine Vision Conference (BMVC), 2018

In order to further enhance the discriminative capability of deep features, we introduce a joint supervision signal, Git loss, which leverages on softmax and center loss functions. The aim of our loss function is to minimize the intra-class variations as well as maximize the inter-class distances.

b3do Learning Inward Scaled Hypersphere Embedding: Exploring Projections in Higher Dimensions
Muhammad Kamran Janjua, Shah Nawaz, Alessandro Calefati, Ignazio Gallo
arXiv preprint arXiv:1810.07037, 2018

We propose to inwardly scale feature representations in proportional to projecting them onto a hypersphere manifold for discriminative analysis.

This super cool design (at least I find it to be cool) is taken from