Stephanus Daniel Handoko

Feasibility Structure Modeling: An Effective Chaperone for Constrained Memetic Algorithms

An important issue in designing memetic algorithms (MAs) is the choice of solutions in the population for local refinements, which becomes particularly crucial when solving computationally expensive problems. With single evaluation of the objective/constraint functions necessitating tremendous computational power and time, it is highly desirable to be able to focus search efforts on the regions where the global optimum is potentially located so as not to waste too many function evaluations. For constrained optimization, the global optimum must either be located at the trough of some feasible basin or some particular point along the feasibility boundary. Presented in this paper is an instance of optinformatics where a new concept of modeling the feasibility structure of inequality-constrained optimization problems-dubbed the feasibility structure modeling-is proposed to perform geometrical predictions of the locations of candidate solutions in the solution space: deep inside any infeasible region, nearby any feasibility boundary, or deep inside any feasible region. This knowledge may be unknown prior to executing an MA but it can be mined as the search for the global optimum progresses. As more solutions are generated and subsequently stored in the database, the feasibility structure can thus be approximated more accurately. As an integral part, a new paradigm of incorporating the classification-rather than the regression-into the framework of MAs is introduced, allowing the MAs to estimate the feasibility boundary such that effective assessments of whether or not the candidate solutions should experience local refinements can be made. This eventually helps preventing the unnecessary refinements and consequently reducing the number of function evaluations required to reach the global optimum.

Extreme learning machine for predicting HLA-Peptide binding

Machine learning techniques have been recognized as powerful tools for learning from data. One of the most popular learning techniques, the Back-Propagation (BP) Artificial Neural Networks, can be used as a computer model to predict peptides binding to the Human Leukocyte Antigens (HLA). The major advantage of computational screening is that it reduces the number of wet-lab experiments that need to be performed, significantly reducing the cost and time. A recently developed method, Extreme Learning Machine (ELM), which has superior properties over BP has been investigated to accomplish such tasks. In our work, we found that the ELM is as good as, if not better than, the BP in term of time complexity, accuracy deviations across experiments, and – most importantly – prevention from over-fitting for prediction of peptide binding to HLA.

QuickVina: Accelerating AutoDock Vina Using Gradient-Based Heuristics for Global Optimization

Predicting binding between macromolecule and small molecule is a crucial phase in the field of rational drug design. AutoDock Vina, one of the most widely used docking software released in 2009, uses an empirical scoring function to evaluate the binding affinity between the molecules and employs the iterated local search global optimizer for global optimization, achieving a significantly improved speed and better accuracy of the binding mode prediction compared its predecessor, AutoDock 4. In this paper, we propose further improvement in the local search algorithm of Vina by heuristically preventing some intermediate points from undergoing local search. Our improved version of Vina-dubbed QVina-achieved a maximum acceleration of about 25 times with the average speed-up of 8.34 times compared to the original Vina when tested on a set of 231 protein-ligand complexes while maintaining the optimal scores mostly identical. Using our heuristics, larger number of different ligands can be quickly screened against a given receptor within the same time frame.

Using classification for constrained memetic algorithm: A new paradigm

Regression has been successfully combined with the memetic algorithm (MA) for constructing surrogate models. It is essentially an attempt to approximate the objective or constraint landscape of a constrained optimization problem. Classification, on the other hand, has probably never been thought of being of any assistance to the MA. In fact, it can be used to approximate the feasibility boundary by means of some decision functions. The search effort can thus be focussed on the nearby region, recalling that many constrained optimization problems have their optimal solutions situated on the boundaries. This simply means that only potential individuals will undergo local refinements, reducing the number of function evaluations and accelerating the identification of the global optimum. Presented in this paper is a new approach that combines the support vector machine (SVM) with the MA to achieve this purpose.

An auction mechanism for the last-mile deliveries via urban consolidation centre

A number of cities around the world have adopted urban consolidation centres (UCCs) to address some challenges of their last-mile deliveries. At the UCC, goods are consolidated based on their destinations prior to their deliveries into the city centre. In many examples, the UCC owns a fleet of eco-friendly vehicles to carry out the deliveries. A carrier/shipper who buys the UCCs service hence no longer needs to enter the city centre where there might be time-window and vehicle-type restrictions. As a result, it becomes possible to retain the use of large trucks for the economies of scale outside the city centre. Furthermore, time which would otherwise be spent in the city centre can then be used to deliver more orders. With possibly tighter regulation and thinning profit margin in near future, requests for the use of the UCCs service shall become more and more common. In this paper, we propose a profit-maximizing auction mechanism for the use of the UCCs service. We first formulate the winner determination problem as mixed-integer program (MIP). Then, we provide a greedy approximation algorithm to solve the MIP in reasonable time. Our experiments indicate that the proposed auction along with the greedy approximation algorithm is able to maximize the UCCs profit to near optimality with reasonable computational budget.

CScore: a simple yet effective scoring function for protein-ligand binding affinity prediction using modified CMAC learning architecture.

Protein-ligand docking is a computational method to identify the binding mode of a ligand and a target protein, and predict the corresponding binding affinity using a scoring function. This method has great value in drug design. After decades of development, scoring functions nowadays typically can identify the true binding mode, but the prediction of binding affinity still remains a major problem. Here we present CScore, a data-driven scoring function using a modified Cerebellar Model Articulation Controller (CMAC) learning architecture, for accurate binding affinity prediction. The performance of CScore in terms of correlation between predicted and experimental binding affinities is benchmarked under different validation approaches. CScore achieves a prediction with R = 0.7668 and RMSE = 1.4540 when tested on an independent dataset. To the best of our knowledge, this result outperforms other scoring functions tested on the same dataset. The performance of CScore varies on different clusters under the leave-cluster-out validation approach, but still achieves competitive result. Lastly, the target-specified CScore achieves an even better result with R = 0.8237 and RMSE = 1.0872, trained on a much smaller but more relevant dataset for each target. The large dataset of protein-ligand complexes structural information and advances of machine learning techniques enable the data-driven approach in binding affinity prediction. CScore is capable of accurate binding affinity prediction. It is also shown that CScore will perform better if sufficient and relevant data is presented. As there is growth of publicly available structural data, further improvement of this scoring scheme can be expected.

Classification-Assisted Memetic Algorithms for Equality-Constrained Optimization Problems

Regressions has successfully been incorporated into memetic algorithm (MA) to build surrogate models for the objective or constraint landscape of optimization problems. This helps to alleviate the needs for expensive fitness function evaluations by performing local refinements on the approximated landscape. Classifications can alternatively be used to assist MA on the choice of individuals that would experience refinements. Support-vector-assisted MA were recently proposed to alleviate needs for function evaluations in the inequality-constrained optimization problems by distinguishing regions of feasible solutions from those of the infeasible ones based on some past solutions such that search efforts can be focussed on some potential regions only. For problems having equality constraints, however, the feasible space would obviously be extremely small. It is thus extremely difficult for the global search component of the MA to produce feasible solutions. Hence, the classification of feasible and infeasible space would become ineffective. In this paper, a novel strategy to overcome such limitation is proposed, particularly for problems having one and only one equality constraint. The raw constraint value of an individual, instead of its feasibility class, is utilized in this work.

Discovering Unique, Low-Energy Transition States Using Evolutionary Molecular Memetic Computing

In the last few decades, identification of transition states has experienced significant growth in research interests from various scientific communities. As per the transition states theory, reaction paths and landscape analysis as well as many thermodynamic properties of biochemical systems can be accurately identified through the transition states. Transition states describe the paths of molecular systems in transiting across stable states. In this article, we present the discovery of unique, low-energy transition states and showcase the efficacy of their identification using the memetic computing paradigm under a Molecular Memetic Computing (MMC) framework. In essence, the MMC is equipped with the tree-based representation of non-cyclic molecules and the covalent-bond-driven evolutionary operators, in addition to the typical backbone of memetic algorithms. Herein, we employ genetic algorithm for the global search, Berny algorithm for individual learning, and make use of the valley-adaptive clearing scheme as the niching strategy in the spirit of Lamarckian learning. Experiments with a number of small non-cyclic molecules demonstrated excellent efficacy of the MMC compared to recent advances of several state-of-the-art algorithms. Not only did the MMC uncover the largest number of transition states, but it also incurred the least amount of computational costs.

A tree-structured covalent-bond-driven molecular memetic algorithm for optimization of ring-deficient molecules

With enormous success in both science and engineering, the recent advances in evolutionary computation-particularly memetic computing-is gaining increasing attention in the molecular optimization community. In this paper, our interest is to introduce a memetic computational methodology for the discovery of low-energy stable conformations-also known as the stereoisomers-of covalently-bonded molecules, due to the abundance of such molecules in nature and their importance in biology and chemistry. To an optimization algorithm, maintaining the same set of bonds over the course of searching for the stereoisomers is a great challenge. Avoiding the steric effect, i.e. preventing atoms from overlapping or getting too close to each other, is another challenge of molecular optimization. Addressing these challenges, three novel nature-inspired tree-based evolutionary operators are first introduced in this paper. A tree-structured covalent-bond-driven molecular memetic algorithm (TCM-MA)-tailored specifically to deal with molecules that involve covalent bonding but contain no cyclic structures using the three novel evolutionary operators-is then proposed for the efficient search of the stereoisomers of ring-deficient covalently-bonded molecules. Through empirical study using the glutamic acid as a sample molecule of interest, it is witnessed that the proposed TCM-MA discovered as many as up to sixteen times more stereoisomers within as little as up to a five times tighter computational budget compared to two other state-of-the-art algorithms.

Achieving Economic and Environmental Sustainabilities in Urban Consolidation Center With Bicriteria Auction

Consolidation lies at the heart of the last-mile logistics problem. Urban consolidation centers (UCCs) have been set up to facilitate such consolidation all over the world. To the best of our knowledge, most-if not all-of the UCCs operate on volume-based fixed-rate charges. To achieve environmental sustainability while ensuring economic sustainability in urban logistics, we propose, in this paper, a bicriteria auction mechanism for the automated assignment of last-mile delivery orders to transport resources. We formulate and solve the winner determination problem of the auction as a biobjective programming model. We then present a systematic way to generate the Pareto frontier to characterize the tradeoff between achieving economic and environmental sustainabilities in urban logistics. Finally, we demonstrate that our proposed bicriteria auction produces the solutions that significantly dominate those obtained from the fixed-rate mechanisms. Our sensitivity analysis on the willingness of carriers to participate in the UCC operation reveals that higher willingness is favorable toward achieving greater good for all, if UCC is designed to be nonprofit and self-sustaining.