Context Aware Image Annotation in Active Learning with Batch Mode

Abstract

Image annotation for active learning is labor-intensive. Various automatic and semi-automatic labeling methods are proposed to save the labeling cost, but a reduction in the number of labeled instances does not guarantee a reduction in cost because the queries that are most valuable to the learner may be the most difficult or ambiguous cases, and therefore the most expensive for an oracle to label accurately. In this paper, we try to solve this problem by using image metadata to offer the oracle more clues about the image during annotation process. We propose a Context Aware Image Annotation Framework (CAIAF) that uses image metadata as similarity metric to cluster images into groups for annotation. We also present useful metadata information as context for each image on the annotation interface. Experiments show that it reduces that annotation cost with CAIAF compared to the conventional framework, while maintaining a high classification performance.

Conversational Structure Aware and Context Sensitive Topic Model for Online Discussions

“…Learning based methods are also explored [52 -54]. In this paper, we will introduce a model that uses the conversational structure [39,40,41] of a discussion thread inherently has to improve the topic modeling performance for short texts within online discussions.…”

Active Learning Based Federated Learning for Waste and Natural Disaster Image Classification

“…More recently, we also analyze the effectiveness of pool-based AL methods in the classification of disaster-related images [11]. Sun et al [12] utilized AL for context-aware image annotation by exploiting the associated additional information available in the form of meta-data. In detail, four different features namely geolocation information, time stamps, users' tags, and camera tags are used in clustering to categorize images into different labeled groups.…”

INSMA: An integrated system for multimodal data acquisition and analysis in the intensive care unit

“…A new dataset is analyzed (Step 4) with the same algorithm used in Step 2, the ANN classifies the output of the algorithm in Step 5, and the result of the ANN classification is then mapped into the physiological phase space in Step 6. This methodology reduces the dimensionality of multiscale variability dynamics in a clinically relevant manner, thereby facilitating the development of clinician-centric visualization tools that can be implemented in a bedside display, and easily integrated in the ICU workflow as a generalized early warning system for clinical decompensation in ICU patients [18]. Any algorithm that quantifies multiscale variability dynamics [16,22] can be used to process the waveform data in order to classify the information extracted from the raw data in an intuitive and physiologically relevant manner [23,24], and thus to facilitate the incorporation of subtle and dynamic fluctuations in physiological waveform data. By assessing the current status of a patient in Fig.…”