A Deep Neural Network for Pairwise Classification: Enabling Feature Conjunctions and Ensuring Symmetry

Atarashi, Kyohei; Oyama, Satoshi; Kurihara, Masahito; Furudo, Kazune

doi:10.1007/978-3-319-57454-7_7

Cited by 11 publications

(12 citation statements)

References 14 publications

(17 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is to find a general linear subspace in which the cumulative pairwise canonical correlation between every pair of feature sets is maximized after the dimension normalization and subspace projection They used multiple feature fusion using subspace learning for face recognition. Atarashi et al [42] used conjunction of features for pairwise classifiers across instances. They then applied the method using support vector machine and simple DNN.…”

Section: B Fusion Of Block Featuresmentioning

confidence: 99%

See 1 more Smart Citation

An Effective Semantic Code Clone Detection Framework Using Pairwise Feature Fusion

2021

View full text Add to dashboard Cite

Code clone detection is important for effective software maintenance. The task is more challenging when clones are semantically similar (Type-IV) in nature, having no structural resemblance to each other. Most existing methods use sequence similarity and/or graph isomorphism between either Abstract Syntax Trees (AST) or Program Dependency Graphs (PDG) to detect Type-I, II and III clones. However, they are mostly unsuccessful in detecting semantic or Type-IV clones. In this work, we propose a novel detection framework using machine learning for automated detection of all four type of clones. The features extracted from a pair code blocks are combined for possible detection of a clone with respect to reference block. We use ASTs and PDGs features of both the code blocks to prepare labelled training samples after fusing the two feature vectors using three different alternatives. We use six (06) state-of-the-art classification models including Deep Convolutional Neural Network to assess the prediction performance of our scheme. To access the effectiveness of our framework we use seven (07) datasets and compare its performance with five state-of-the-art clone detectors. We also compare a large number of algorithms for code clone detection. Comparing the performance of a large number of machine learning techniques, ANN and non-ANN, using such features, and establishing that fusing of AST and PDG features gives competitive results using deep learning as well as boosted tree algorithms, and find that boosted tree algorithms like XGBOOT are quite competitive in clone detection. Experimental results demonstrate that our approach outperforms existing clone detection methods in terms of prediction accuracy.

show abstract

Section: B Fusion Of Block Featuresmentioning

confidence: 99%

“…Fu et al [41] Face recognition Multiple Feature Fusion. Atarashi et al [42] Entity resolution and symmetry Linear, multiplicative, and distance combinations. Lu et at.…”

Section: Volume 4 2016mentioning

confidence: 99%

An Effective Semantic Code Clone Detection Framework Using Pairwise Feature Fusion

2021

View full text Add to dashboard Cite

show abstract

“…While the DNN can extract useful feature representation, introducing the layer using feature combinations explicitly improves the performance of the DNN. Indeed, the DNN-based models using feature combinations outperformed simple DNNs in some applications [23], [24], [26], [27]. We also propose a higher-order pairwise deep neural network (HOPairDNN) by defining z (m) s = m t=2 P t (u (t) s , u (t) s ), (a, b) for s ∈ [k] in Eq.…”

Section: Higher-order Pairwise Deep Neural Networkmentioning

confidence: 99%

“…• DBLP. For the author disambiguation task, we extracted 3,384 papers in which there were 729 unique author names from the DBLP dataset [23]. Each paper was considered an object.…”

Section: Datasetsmentioning

confidence: 99%

See 1 more Smart Citation

Link Prediction Using Higher-Order Feature Combinations across Objects

Atarashi

Oyama

Kurihara

2020

IEICE Trans. Inf. & Syst.

Self Cite

View full text Add to dashboard Cite

Link prediction, the computational problem of determining whether there is a link between two objects, is important in machine learning and data mining. Feature-based link prediction, in which the feature vectors of the two objects are given, is of particular interest because it can also be used for various identification-related problems. Although the factorization machine and the higher-order factorization machine (HOFM) are widely used for feature-based link prediction, they use feature combinations not only across the two objects but also from the same object. Feature combinations from the same object are irrelevant to major link prediction problems such as predicting identity because using them increases computational cost and degrades accuracy. In this paper, we present novel models that use higher-order feature combinations only across the two objects. Since there were no algorithms for efficiently computing higher-order feature combinations only across two objects, we derive one by leveraging reported and newly obtained results of calculating the ANOVA kernel. We present an efficient coordinate descent algorithm for proposed models. We also improve the effectiveness of the existing one for the HOFM. Furthermore, we extend proposed models to a deep neural network. Experimental results demonstrated the effectiveness of our proposed models.

show abstract

A complete graph-based approach with multi-task learning for predicting synergistic drug combinations

Wang,

Zhu,

Chen

et al. 2023

Bioinformatics

View full text Add to dashboard Cite

Motivation Drug combination therapy shows significant advantages over monotherapy in cancer treatment. Since the combinational space is difficult to be traversed experimentally, identifying novel synergistic drug combinations based on computational methods has become a powerful tool for pre-screening. Among them, methods based on deep learning have far outperformed other methods. However, most deep learning-based methods are unstable and will give inconsistent predictions even by simply changing the input order of drugs. In addition, the insufficient experimental data of drug combination screening limits the generalization ability of existing models. These problems prevent the deep learning-based models from being in service. Results In this paper, we propose CGMS to address the above problems. CGMS models a drug combination and a cell line as a heterogeneous complete graph, and generates the whole-graph embedding to characterize their interaction by leveraging the heterogeneous graph attention network. Based on the whole-graph embedding, CGMS can make a stable, order-independent prediction. To enhance the generalization ability of CGMS, we apply the multi-task learning technique to train the model on drug synergy prediction task and drug sensitivity prediction task simultaneously. We compare CGMS’s generalization ability with 6 state-of-the-art methods on a public dataset, and CGMS significantly outperforms other methods in the leave-drug combination-out scenario, as well as in the leave-cell line-out and leave-drug-out scenarios. We further present the benefit of eliminating the order dependency and the discrimination power of whole-graph embeddings, interpret the rationality of the attention mechanism, and verify the contribution of multi-task learning. Availability The code of CGMS is available via https://github.com/TOJSSE-iData/CGMS Supplementary information Supplementary data are available at Bioinformatics

show abstract

A Deep Neural Network for Pairwise Classification: Enabling Feature Conjunctions and Ensuring Symmetry

Cited by 11 publications

References 14 publications

An Effective Semantic Code Clone Detection Framework Using Pairwise Feature Fusion

An Effective Semantic Code Clone Detection Framework Using Pairwise Feature Fusion

Link Prediction Using Higher-Order Feature Combinations across Objects

A complete graph-based approach with multi-task learning for predicting synergistic drug combinations

Contact Info

Product

Resources

About