Tuan Pham Minh

Tuan Pham Minh, Jayan Wijesingha, Daniel Kottke et al.

Remote sensing through semantic segmentation of satellite images contributes to the understanding and utilisation of the earth's surface. For this purpose, semantic segmentation networks are typically trained on large sets of labelled satellite images. However, obtaining expert labels for these images is costly. Therefore, we propose to rely on a low-cost approach, e.g. crowdsourcing or pretrained networks, to label the images in the first step. Since these initial labels are partially erroneous, we use active learning strategies to cost-efficiently refine the labels in the second step. We evaluate the active learning strategies using satellite images of Bengaluru in India, labelled with land cover and land use labels. Our experimental results suggest that an active label refinement to improve the semantic segmentation network's performance is beneficial.

Daniel Kottke, Georg Krempl, Marianne Stecklina et al.

In machine learning, active class selection (ACS) algorithms aim to actively select a class and ask the oracle to provide an instance for that class to optimize a classifier's performance while minimizing the number of requests. In this paper, we propose a new algorithm (PAL-ACS) that transforms the ACS problem into an active learning task by introducing pseudo instances. These are used to estimate the usefulness of an upcoming instance for each class using the performance gain model from probabilistic active learning. Our experimental evaluation (on synthetic and real data) shows the advantages of our algorithm compared to state-of-the-art algorithms. It effectively prefers the sampling of difficult classes and thereby improves the classification performance.

Kien Duy Nguyen, Tuan Pham Minh, Quang Nhat Nguyen et al.

Friend suggestion is a fundamental problem in social networks with the goal of assisting users in creating more relationships, and thereby enhances interest of users to the social networks. This problem is often considered to be the link prediction problem in the network. ZingMe is one of the largest social networks in Vietnam. In this paper, we analyze the current approach for the friend suggestion problem in ZingMe, showing its limitations and disadvantages. We propose a new efficient approach for friend suggestion that uses information from the network structure, attributes and interactions of users to create resources for the evaluation of friend connection amongst users. Friend connection is evaluated exploiting both direct communication between the users and information from other ones in the network. The proposed approach has been implemented in a new system version of ZingMe. We conducted experiments, exploiting a dataset derived from the users' real use of ZingMe, to compare the newly proposed approach to the current approach and some well-known ones for the accuracy of friend suggestion. The experimental results show that the newly proposed approach outperforms the current one, i.e., by an increase of 7% to 98% on average in the friend suggestion accuracy. The proposed approach also outperforms other ones for users who have a small number of friends with improvements from 20% to 85% on average. In this paper, we also discuss a number of open issues and possible improvements for the proposed approach.