Back

Speaker "Humayun Irshad" Details Back

 

Topic

An Active Learning Framework to Optimize Training of Deep Model with Human-in-the-loop

Abstract

Deep learning models have been used extensively to solve real-world problems in recent years. The performance of such models relies heavily on large amounts of labeled data for training. While the advances of data collection technology have enabled the acquisition of a massive volume of data, labeling the data remains an expensive and time-consuming task. Random selection of data points for generating training data may lead to a time consuming and inefficient process particularly when there is a high variation in the scale, shape and orientation of the data, or when sample data points do not follow an even distribution. Active learning techniques are being progressively adopted to accelerate the development of machine learning solutions by allowing the model to query the data they learn from. In this talk, an active learning framework with transfer learning and crowd sourcing approach is introduced to solve a real-world problem in transportation and autonomous driving discipline, parking sign recognition, for which a large amount of unlabeled data is available and proposed an active learning-based approach to address it. The main novelty of proposed framework is a crowd sourcing based active learning framework which intelligently select a small subset of images that efficiently improved the performance of object detection model. For the iterative part of the proposed active learning framework, we defined a criteria for the selection of images which are good candidates for building a better model. These candidate images were chosen to cover not only a diverse range of parking signs but also challenging corner cases where the parking signs are either partially occluded, blurred, reflecting light or are placed far away in the background. We discuss how such a framework contributes to building an accurate model in a cost-effective and fast way to solve the parking sign recognition problem in spite of the unevenness of the data associated with the fact that street-level images (such as parking signs) vary in shape, color, orientation and scale, and often appear on top of different types of background. In this talk, an active learning framework with crowd sourcing approach is introduced to solve a real-world problem in transportation and autonomous driving discipline, parking sign recognition, for which a large amount of unlabeled data is available. It generates an accurate model in a cost-effective and fast way to solve the parking sign recognition problem in spite of the unevenness of the data.

Profile

Humayun Irshad is currently the Lead Scientist of Machine Learning at Figure Eight, the essential human-in-the-loop AI platform for data science and machine learning teams. He has expertise in developing machine learning, more specifically deep learning frameworks for various applications like object detection, segmentation and classification in fields ranging from medical, retail, self-driving car, satellite, fashion, etc. Now a days, he is building Active Learning frameworks for selection of training data from labeled or unlabeled dataset to build model to avoid over-training and dealing corner cases. He has 3 years PostDoc experience at Harvard Medical School where he developed machine learning and deep learning frameworks for Computer Aided Diagnosis system including region of interest detection, nuclei and gland detection, segmentation and classification in 2D and 3D medical images. He got a PhD in Computer Science from University of Grenoble France.