Last Publications

2021
Charles A. Kantor et al. 2021. “Gradient-Based Localization and Spatial Attention for Confidence Measure in Fine-Grained Recognition using Deep Neural Networks.” Proc. 35th AAAI Conf. on Artificial Intelligence (AAAI'21).
Charles A. Kantor et al. 2021. “Over-MAP: Structural Attention Mechanism and Automated Semantic Segmentation Ensembled for Uncertainty Prediction.” Proc. 33rd Annual Conference on Innovative Applications of Artificial Intelligence (IAAI'­21).Abstract

Both theoretical and practical problems in deep learning classification require solutions for assessing uncertainty prediction but current state-of-the-art methods in this area are computationally expensive. In this paper, we propose a new confidence measure dubbed Over-MAP that utilizes a measure of overlap between structural attention mechanisms and segmentation methods, that is of particular interest in accurate fine-grained contexts. We show that this classification confidence increases with the degree of overlap. The associated confidence and identification tools are conceptually simple, efficient, and of high practical interest as they allow for weeding out misleading examples in training data. Our measure is currently deployed in the real-world on widely used platforms to annotate large-scale data efficiently.

Copyright by the authors. All rights reserved to authors only. Correspondence to: ckantor (at) stanford [dot] edu

2020
Charles A. Kantor et al. 8/2020. “Geo-Spatiotemporal Features and Shape-Based Prior Knowledge for Fine-grained Imbalanced Data Classification.” Proc. IJCAI 2021 Workshop on AI for Social Good (2021). Japan. Publisher's VersionAbstract

Fine-grained classification aims at distinguishing between items with similar global perception and patterns, but that differ by minute details. Our primary challenges comes from both small inter-class variations and large intra-class variations. In this article, we propose to combine several innovations to improve fine-grained classification within the use-case of fauna, which is of practical interest for experts. We utilize geo-spatiotemporal data to enrich the picture information and further improve the performance. We also investigate state-of-the-art methods for handling the imbalanced data issue.

Copyright by the authors. All rights reserved to authors only. Correspondence to: ckantor (at) stanford [dot] edu

2103.11285.pdf
Charles A. Kantor et al. 2020. “Asymptotic Cross-Entropy Weighting and Guided-Loss in Supervised Hierarchical Setting using Deep Attention Networks.” AAAI Fall Symposium on AI for Social Good (AI4SG).Abstract

This article reveals two main techniques for improving fine-grained recognition and classification, defined as executing these tasks between items with similar general patterns but that differ through small details. First, we build a preliminary automated segmentation algorithm to ignore the image's background for our guided-classification through the introduction of our attention-based loss. To do this, we wield segmentation-issued maps to make the classification network's training easier through an additional loss that penalizes attention given to features outside the mask using convolutional block. Furthermore, we proffer a hierarchical loss based on cross-entropy penalizing parent-level classification to leverage each wildlife species' philology. We applied our approaches in the particular context of wildlife recognition and analytics.

Copyright by the authors. All rights reserved to authors only. Correspondence to: ckantor (at) stanford [dot] edu