COIN / Publications

Verified from Google Scholar · 10 August 2026

The COIN publication lineage—from representation to real problems.

This map separates direct Coincidence Algorithm papers from methodological ancestors and closely related applications. It makes the boundary visible instead of labelling every optimization paper as COIN.

Editorial method. The bibliographic selection was checked against Warin Wattanapornprom’s Google Scholar profile sorted by publication date. The paragraphs below are reconstructed abstracts: new summaries written from verified publication metadata, available manuscripts, dissertation records and the stated research scope. They are not quotations and must not be cited as the publishers’ original abstracts. Follow “Scholar record” for the source record.
Direct COINMethod lineageRelated application

Foundation · 2009–2013

Defining COIN, negative knowledge and multimodal search

2009 · Direct COIN · IEEE CEC

Multi-objective combinatorial optimisation with coincidence algorithm

W. Wattanapornprom, P. Olanviwitchai, P. Chutima, and P. Chongstitvatana · pp. 1675–1682

Reconstructed abstract

This paper extends Coincidence Algorithm learning to permutation problems with more than one objective. Rather than collapsing every criterion into one scalar, candidate quality is interpreted through nondominance so the learned model can sample multiple trade-off regions. The work establishes the early multi-objective COIN formulation and examines whether probability learned from partial order can support both convergence and distribution across a Pareto set.

Scholar record ↗
2010 · Direct COIN · Electrical Engineering Conference

Coincidence Algorithm for Combinatorial Optimisation and Its Applications

P. Chongstitvatana, W. Wattanapornprom, P. Olanviwitchai, R. Sirovetnukul, et al.

Reconstructed abstract

This overview presents COIN as a probabilistic optimizer for permutation-encoded combinatorial problems. Candidate solutions contribute incidences to a learned model, and the model constructs later populations without requiring a conventional crossover operator. The paper connects the common mechanism to several applications and positions COIN as a reusable family rather than a single problem-specific solver.

Scholar record ↗
2010 · Doctoral dissertation · Chulalongkorn University

Hybrid positive and negative correlation learning in estimation of distribution algorithm for combinatorial optimization problems

W. Wattanapornprom

Reconstructed abstract

The dissertation studies how an estimation-of-distribution search can learn from both strong and weak permutations. Positive evidence reinforces structures associated with good candidates; negative evidence suppresses recurrent structures associated with poor ones. Controlled multimodal puzzles and applied permutation problems are used to examine solution quality, quantity and diversity. The resulting architecture provides the conceptual foundation for COIN’s reward–punishment learning and its later scalar and multi-objective variants.

Scholar record ↗
2011 · Method lineage · IEEE IEEM

The effectiveness of hybrid negative correlation learning in evolutionary algorithm for combinatorial optimization problems

R. Sirovetnukul, P. Chutima, W. Wattanapornprom, and P. Chongstitvatana

Reconstructed abstract

This study isolates hybrid negative-correlation learning as a diversity mechanism for combinatorial evolutionary search. Multiple learners are encouraged to avoid duplicating the same unproductive structures while still exploiting useful evidence. The experiments ask when combining positive and negative signals improves exploration and when conflicting evidence can weaken learning, making the paper an important bridge between the dissertation hypothesis and reusable COIN selection policies.

Scholar record ↗
2011 · Method lineage · GECCO Companion

Solving multimodal combinatorial puzzles with edge-based estimation of distribution algorithm

W. Wattanapornprom and P. Chongstitvatana

Reconstructed abstract

The paper uses an edge-based probability model to search combinatorial puzzles with several distinct high-quality answers. Adjacency relationships provide a representation that can be shared across solutions even when absolute positions differ. The multimodal setting shifts evaluation beyond a single best score toward the number and diversity of optima recovered, creating the experimental foundation for later COIN work on negative knowledge and multiple modes.

Scholar record ↗
2012 · Direct COIN · JCSSE

Solving multimodal problems by coincidence algorithm

K. Waiyapara and P. Chongstitvatana

Reconstructed abstract

This work applies COIN directly to multimodal optimization, where success means retaining several promising solution families instead of converging to only one. Coincidence statistics capture structures shared within and across candidate modes, while learned suppression reduces repeated weak combinations. The paper contributes evidence that COIN’s model can be used to preserve alternative optima in discrete search spaces.

Scholar record ↗
2013 · Direct COIN · IEICE Transactions

Negative Correlation Learning in the Estimation of Distribution Algorithms for Combinatorial Optimization

W. Wattanapornprom and P. Chongstitvatana · E96-D(11), 2397–2408

Reconstructed abstract

This journal paper formalizes the use of negative correlation in estimation-of-distribution algorithms for permutation search. Instead of allowing probability models to repeat the same structures, learning pressure encourages complementary exploration while preserving evidence from good solutions. The study evaluates how the mechanism affects premature convergence, discovery of multiple solutions and search behavior across combinatorial problems, providing the clearest archival account of negative knowledge in the COIN lineage.

Scholar record ↗

Direct applications · 2013–2016

Changing the representation without abandoning the learning idea

2013 · Direct COIN · JCSSE

Solving Sudoku puzzles with node based Coincidence algorithm

K. Waiyapara, W. Wattanapornprom, and P. Chongstitvatana

Reconstructed abstract

Sudoku is encoded as a permutation-oriented search problem and solved with node-based COIN, where probabilities associate values with positions rather than only adjacent edges. The work demonstrates how representation can enforce part of the puzzle structure before evaluation, leaving COIN to learn which remaining assignments reduce row, column and block conflicts. It provides the historical baseline for the current Sudoku laboratory and its newer prime-product and decomposed objectives.

Scholar record ↗
2013 · Direct COIN · JCSSE

Application of Node Based Coincidence algorithm for flow shop scheduling problems

O. Srimongkolkul and P. Chongstitvatana · pp. 49–52

Reconstructed abstract

This paper applies NB-COIN to permutation flow-shop scheduling with total flow time as the objective. A node-position probability model learns where jobs tend to occur in stronger schedules while positive and negative samples update the distribution. Experiments on Taillard instances compare solution quality and computational demand with established methods; the paper reports solutions within 1.7% of then-recent best-known values while emphasizing modest computational resources.

Scholar record ↗
2016 · Direct COIN · ISCIT

RNA secondary structure prediction with coincidence algorithm

S. Srikamdee, W. Wattanapornprom, and P. Chongstitvatana

Reconstructed abstract

RNA secondary-structure prediction is formulated as selection of a compatible helix subset whose estimated free energy is minimized under an INN-HB model. COIN samples permutations of candidate helices, accepting compatible structures during decoding to reduce repair cost. Ten known RNA sequences of different lengths and organisms are evaluated by sensitivity, specificity and F-measure against RNApredict and SARNA-Predict; the paper reports higher average values for COIN across all three measures.

Scholar record ↗

Industrial lineage · 2013–2015

Order acceptance, capacity and production decisions

2013 · Related EDA application · IEEE IEEM

Application of estimation of distribution algorithms for solving order acceptance with weighted tardiness problems

W. Wattanapornprom, T. Li, W. Wattanapornprom, and P. Chongstitvatana

Reconstructed abstract

This study joins order selection with production sequencing when accepted work can incur weighted tardiness. An estimation-of-distribution search represents candidate decisions and learns patterns associated with profitable, feasible schedules. The problem broadens the COIN research context from benchmark permutations to coupled commercial decisions: which orders to accept, how to arrange them, and how lateness changes the value of a schedule.

Scholar record ↗
2013 · Industrial context · IEEE IEEM

The merging of MPS and order acceptance in a semi-order-driven industry: A case study of the parasol industry

W. Wattanapornprom and T. Li

Reconstructed abstract

This case study connects master production scheduling with order-acceptance decisions in a semi-order-driven parasol business. It clarifies the operational setting that later node-based optimization papers address: finite capacity, uncertain order combinations and the need to balance utilization against delivery performance. The work is included as application context, not as a direct COIN algorithm paper.

Scholar record ↗
2014 · Node-EDA lineage · ICSEC

Application of node based estimation of distribution algorithms for solving order acceptance with capacity balancing problems by trading off between over capacity and under capacity

W. Wattanapornprom, T. Li, W. Wattanapornprom, and P. Chongstitvatana · pp. 238–243

Reconstructed abstract

The paper models order acceptance as a trade-off between unused capacity and overload across a production plan. A node-based probability model learns assignment and ordering patterns from evaluated candidates. This representation is closely related to NB-COIN but is described in the publication as a node-based EDA; it extends the method lineage toward decision problems whose objectives reflect opposing operational risks.

Scholar record ↗
2014 · Node-EDA lineage · IEEE IEEM

Overtime capacity expansion in order acceptance with node based estimation of distribution algorithms

W. Wattanapornprom, T. Li, W. Wattanapornprom, and P. Chongstitvatana

Reconstructed abstract

This work adds overtime as a controllable capacity-expansion decision to order acceptance. The optimizer must compare additional revenue from accepted orders with overtime and capacity consequences, while a node-based distribution learns recurring structures in better plans. The study shows how the representation lineage can incorporate business actions instead of optimizing a fixed schedule alone.

Scholar record ↗
2015 · Direct node-based COIN · Manufacturing proceedings

Application of node based coincidence algorithm for solving order acceptance with multi-process capacity balancing problems

W. Wattanapornprom, T. Li, W. Wattanapornprom, and P. Chongstitvatana

Reconstructed abstract

This paper applies node-based COIN to order acceptance when several linked processes impose separate capacity constraints. Candidate plans must coordinate selection and allocation across stages rather than balancing one aggregate resource. The learned node distribution guides construction toward combinations that use capacity coherently, extending direct COIN application from sequencing benchmarks to multi-process production decisions.

Scholar record ↗
2015 · Optimization context · Industrial Engineering Journal

An Optimization Model and Algorithms for Order Acceptance Problem with Overtime Strategy

W. Wattanapornprom, L. Tie-ke, and W. Bai-lin · 18(2), 1–08

Reconstructed abstract

This article formalizes an order-acceptance model in which overtime can expand production capacity at a cost. The decision framework evaluates accepted demand, capacity usage and overtime strategy together, providing a business-oriented objective and constraint system for heuristic solvers. It is included as related application context for the node-based COIN/EDA studies rather than labelled as a direct COIN contribution.

Scholar record ↗

Use the map as a starting point, not a substitute for the papers.

For claims about algorithms, datasets or results, cite and read the original publication. For the current implementation, also cite the COINCIDENCE Algorithms Suite release and report the exact commit, evaluator, parameters, seeds and evaluation budget.