Metabolic networks classification and knowledge discovery by information granulation

Recent discretization-based feature selection methods show great advantages by introducing the entropy-based cut-points for features to integrate discretization and feature selection into one stage for high-dimensional data. However, current methods usually consider the individual features independently, ignoring the interaction between features with cut-points and those without cut-points, which results in information loss. In this paper, we propose a cooperative coevolutionary algorithm based on the genetic algorithm (GA) and particle swarm optimization (PSO), which searches for the feature subsets with and without entropy-based cut-points simultaneously. For the features with cut-points, a ranking mechanism is used to control the probability of mutation and crossover in GA. In addition, a binary-coded PSO is applied to update the indices of the selected features without cut-points. Experimental results on 10 real datasets verify the effectiveness of our algorithm in classification accuracy compared with several state-of-the-art competitors.

“…Considering the above issues, we use the

[ 32 ] and

where

is a weight coefficient to combine the

and

.…”

Section: Proposed Methodsmentioning

confidence: 99%

A Cooperative Coevolutionary Approach to Discretization-Based Feature Selection for High-Dimensional Data

Kang

Zhang

2020

“…Embedding-based representation methods are also used for textual data, and there are several effective embedding methods such as Skim-gram [ 14 ], latent semantic indexing (LSI) [ 15 ], latent Dirichlet allocation (LDA) [ 16 ], as well as some variants of them in [ 17 , 18 , 19 ]. Granular Computing paradigm [ 20 , 21 , 22 ] is an embedding method which is powerful especially when dealing with non-conventional data such as graphs, sequences, text documents. However, the embedding representation for textual data is significantly different from categorical data since categorical data is structured, whereas textual data is unstructured.…”

Section: Related Workmentioning

confidence: 99%

CDE++: Learning Categorical Data Embedding by Enhancing Heterogeneous Feature Value Coupling Relationships

Dong

Jian

Zuo

2020

Categorical data are ubiquitous in machine learning tasks, and the representation of categorical data plays an important role in the learning performance. The heterogeneous coupling relationships between features and feature values reflect the characteristics of the real-world categorical data which need to be captured in the representations. The paper proposes an enhanced categorical data embedding method, i.e., CDE++, which captures the heterogeneous feature value coupling relationships into the representations. Based on information theory and the hierarchical couplings defined in our previous work CDE (Categorical Data Embedding by learning hierarchical value coupling), CDE++ adopts mutual information and margin entropy to capture feature couplings and designs a hybrid clustering strategy to capture multiple types of feature value clusters. Moreover, Autoencoder is used to learn non-linear couplings between features and value clusters. The categorical data embeddings generated by CDE++ are low-dimensional numerical vectors which are directly applied to clustering and classification and achieve the best performance comparing with other categorical representation learning methods. Parameter sensitivity and scalability tests are also conducted to demonstrate the superiority of CDE++.

“…In order to deal with such domains, five mainstream approaches can be pursued [ 10 ]: Feature generation and/or feature engineering, where numerical features are extracted ad-hoc from structured patterns (e.g., using their properties or via measurements) and can be further merged according to different strategies (e.g., in a multi-modal way [ 11 ]); Ad-hoc dissimilarities in the input space, where custom dissimilarity measures are designed in order to process structured patterns directly in the input domain without moving towards Euclidean (or metric) spaces. Common—possibly parametric—edit distances include the Levenshtein distance [ 12 ] for sequence domains and graph edit distances [ 13 ] for graphs domains; Embedding via information granulation and granular computing [ 3 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 ]; Dissimilarity representations [ 26 , 27 , 28 ], where structured patterns are embedded in the Euclidean space according to their pairwise dissimilarities; Kernel methods, where the mapping between the original input space and the Euclidean space exploits positive-definite kernel functions [ 29 , 30 , 31 , 32 , 33 ]. …”

Section: Introductionmentioning

confidence: 99%

Modelling and Recognition of Protein Contact Networks by Multiple Kernel Learning and Dissimilarity Representations

Martino

Santis

Giuliani

2020

Self Cite

Multiple kernel learning is a paradigm which employs a properly constructed chain of kernel functions able to simultaneously analyse different data or different representations of the same data. In this paper, we propose an hybrid classification system based on a linear combination of multiple kernels defined over multiple dissimilarity spaces. The core of the training procedure is the joint optimisation of kernel weights and representatives selection in the dissimilarity spaces. This equips the system with a two-fold knowledge discovery phase: by analysing the weights, it is possible to check which representations are more suitable for solving the classification problem, whereas the pivotal patterns selected as representatives can give further insights on the modelled system, possibly with the help of field-experts. The proposed classification system is tested on real proteomic data in order to predict proteins’ functional role starting from their folded structure: specifically, a set of eight representations are drawn from the graph-based protein folded description. The proposed multiple kernel-based system has also been benchmarked against a clustering-based classification system also able to exploit multiple dissimilarities simultaneously. Computational results show remarkable classification capabilities and the knowledge discovery analysis is in line with current biological knowledge, suggesting the reliability of the proposed system.