arXiv subjects

Jun Huan

Publications and source records attributed to Jun Huan.

42 records · Page 3Linked to original sources

arXiv2018

Data Dropout: Optimizing Training Data for Convolutional Neural Networks

Deep learning models learn to fit training data while they are highly expected to generalize well to testing data. Most works aim at finding such models by creatively designing architectures and fine-tuning parameters. To adapt to particular tasks, hand-crafted information such as image prior has also been incorporated into end-to-end learning. However, very little progress has been made on investigating how an individual training sample will influence the generalization ability of a model. In other words, to achieve high generalization accuracy, do we really need all the samples in a training dataset? In this paper, we demonstrate that deep learning models such as convolutional neural networks may not favor all training samples, and generalization accuracy can be further improved by dropping those unfavorable samples. Specifically, the influence of removing a training sample is quantifiable, and we propose a Two-Round Training approach, aiming to achieve higher generalization accuracy. We locate unfavorable samples after the first round of training, and then retrain the model from scratch with the reduced training dataset in the second round. Since our approach is essentially different from fine-tuning or further training, the computational cost should not be a concern. Our extensive experimental results indicate that, with identical settings, the proposed approach can boost performance of the well-known networks on both high-level computer vision problems such as image classification, and low-level vision problems such as image denoising.

cs.CV↗

arXiv2017

Fast Computation of Graph Edit Distance

The graph edit distance (GED) is a well-established distance measure widely used in many applications. However, existing methods for the GED computation suffer from several drawbacks including oversized search space, huge memory consumption, and lots of expensive backtracking. In this paper, we present BSS_GED, a novel vertex-based mapping method for the GED computation. First, we create a small search space by reducing the number of invalid and redundant mappings involved in the GED computation. Then, we utilize beam-stack search combined with two heuristics to efficiently compute GED, achieving a flexible trade-off between available memory and expensive backtracking. Extensive experiments demonstrate that BSS GED is highly efficient for the GED computation on sparse as well as dense graphs and outperforms the state-of-the-art GED methods. In addition, we also apply BSS_GED to the graph similarity search problem and the practical results confirm its efficiency.

cs.DS↗

arXiv2017

On the Unreported-Profile-is-Negative Assumption for Predictive Cheminformatics

In cheminformatics, compound-target binding profiles has been a main source of data for research. For data repositories that only provide positive profiles, a popular assumption is that unreported profiles are all negative. In this paper, we caution audience not to take this assumption for granted, and present empirical evidence of its ineffectiveness from a machine learning perspective. Our examination is based on a setting where binding profiles are used as features to train predictive models; we show (1) prediction performance degrades when the assumption fails and (2) explicit recovery of unreported profiles improves prediction performance. In particular, we propose a framework that jointly recovers profiles and learns predictive model, and show it achieves further performance improvement. The presented study not only suggests applying matrix recovery methods to recover unreported profiles, but also initiates a new missing feature problem which we called Learning with Positive and Unknown Features.

cs.LG↗

arXiv2017

Discriminatory Transfer

We observe standard transfer learning can improve prediction accuracies of target tasks at the cost of lowering their prediction fairness -- a phenomenon we named discriminatory transfer. We examine prediction fairness of a standard hypothesis transfer algorithm and a standard multi-task learning algorithm, and show they both suffer discriminatory transfer on the real-world Communities and Crime data set. The presented case study introduces an interaction between fairness and transfer learning, as an extension of existing fairness studies that focus on single task learning.

cs.CY↗

arXiv2016

MSQ-Index: A Succinct Index for Fast Graph Similarity Search

Graph similarity search has received considerable attention in many applications, such as bioinformatics, data mining, pattern recognition, and social networks. Existing methods for this problem have limited scalability because of the huge amount of memory they consume when handling very large graph databases with millions or billions of graphs. In this paper, we study the problem of graph similarity search under the graph edit distance constraint. We present a space-efficient index structure based upon the q-gram tree that incorporates succinct data structures and hybrid encoding to achieve improved query time performance with minimal space usage. Specifically, the space usage of our index requires only 5%-15% of the previous state-of-the-art indexing size on the tested data while at the same time achieving 2-3 times acceleration in query time with small data sets. We also boost the query performance by augmenting the global filter with range search, which allows us to perform a query in a reduced region. In addition, we propose two effective filters that combine degree structures and label structures. Extensive experiments demonstrate that our proposed approach is superior in space and competitive in filtering to the state-of-the-art approaches. To the best of our knowledge, our index is the first in-memory index for this problem that successfully scales to cope with the large dataset of 25 million chemical structure graphs from the PubChem dataset.

cs.DB↗

arXiv2016

Learning Social Circles in Ego Networks based on Multi-View Social Graphs

In social network analysis, automatic social circle detection in ego-networks is becoming a fundamental and important task, with many potential applications such as user privacy protection or interest group recommendation. So far, most studies have focused on addressing two questions, namely, how to detect overlapping circles and how to detect circles using a combination of network structure and network node attributes. This paper asks an orthogonal research question, that is, how to detect circles based on network structures that are (usually) described by multiple views? Our investigation begins with crawling ego-networks from Twitter and employing classic techniques to model their structures by six views, including user relationships, user interactions and user content. We then apply both standard and our modified multi-view spectral clustering techniques to detect social circles in these ego-networks. Based on extensive automatic and manual experimental evaluations, we deliver two major findings: first, multi-view clustering techniques perform better than common single-view clustering techniques, which only use one view or naively integrate all views for detection, second, the standard multi-view clustering technique is less robust than our modified technique, which selectively transfers information across views based on an assumption that sparse network structures are (potentially) incomplete. In particular, the second finding makes us believe a direct application of standard clustering on potentially incomplete networks may yield biased results. We lightly examine this issue in theory, where we derive an upper bound for such bias by integrating theories of spectral clustering and matrix perturbation, and discuss how it may be affected by several network characteristics.

cs.SI↗