Pembentukan Solusi Awal International Timetabling Competition 2021

Penjadawalan olahraga merupakan salah satu cabang dari optimasi di riset operasi. Penjadwalan olahraga memiliki berbagai macam batasan yang menantang para peneliti untuk menyelesaikannya. International Timetabling Competition (ITC) 2021 merupakan salah satu kompetisi optimasi yang menyediakan permasalahan penjadwalan olahraga. Permasalahan utama pada ITC 2021 yaitu menentukan jadwal waktu yang tepat untuk sebuah pertandingan. Sebuah jadwal dikatakan dapat digunakan (feasible) apabila tidak melanggar hard constraint yang ada. Pembentukan solusi awal yang feasible saat ini dapat dilakukan dengan algoritma constraint programming atau integer programming. Akan tetapi, kedua algoritma tersebut cukup rumit untuk diimplementasikan. Penelitian ini berfokus pada pembentukan solusi awal yang feasible dengan cara yang mudah untuk diimplementasikan. Cara yang digunakan yaitu dengan mengoptimasi pelanggaran hard constraint menggunakan algoritma Late Acceptance Hill Climbing (LAHC) dan Tabu Search dengan kerangka kerja Hyper-Heuristic yang melibatkan low level heuristic (LLH). Algoritma dijalankan maksimal dengan batasan waktu 6 jam untuk setiap data. Hasil dari optimasi pelanggaran hard constraint menggunakan algoritma LAHC dan tabu search dapat menghasilkan solusi awal yang feasible sebanyak 44.44% atau 24 dari 54 keseluruhan dataset.

Constrained versus Unconstrained Rational Inattention

The rational inattention literature is split between two versions of the model: in one, mutual information of states and signals are bounded by a hard constraint, while, in the other, it appears as an additive term in the decision maker’s utility function. The resulting constrained and unconstrained maximization problems are closely related, but, nevertheless, their solutions differ in certain aspects. In particular, movements in the decision maker’s prior belief and utility function lead to opposite comparative statics conclusions.

Ant colony optimization with semi random initialization for nurse rostering problem

A nurse rostering problem is an NP-Hard problem that is difficult to solve during the complexity of the problem. Since good scheduling is the schedule that fulfilled the hard constraint and minimizes the violation of soft constraint, a lot of approaches is implemented to improve the quality of the schedule. This research proposed an improvement on ant colony optimization with semi-random initialization for nurse rostering problems. Semi-random initialization is applied to avoid violation of the hard constraint, and then the violation of soft constraint will be minimized using ant colony optimization. Semi-random initialization will improve the construction solution phase by assigning nurses directly to the shift that is related to the hard constraint, so the violation of hard constraint will be avoided from the beginning part. The scheduling process will complete by pheromone value by giving weight to the rest available shift during the ant colony optimization process. This proposed method is tested using a real-world problem taken from St. General Hospital Elisabeth. The objective function is formulated to minimize the violation of the constraints and balance nurse workload. The performance of the proposed method is examined by using different dimension problems, with the same number of ant and iteration. The proposed method is also compared to conventional ant colony optimization and genetic algorithm for performance comparison. The experiment result shows that the proposed method performs better with small to medium dimension problems. The semi-random initialization is a success to avoid violation of the hard constraint and minimize the objective value by about 24%. The proposed method gets the lowest objective value with 0,76 compared to conventional ant colony optimization with 124 and genetic algorithm with 1.

Differentiable Subset Pruning of Transformer Heads

Multi-head attention, a collection of several attention mechanisms that independently attend to different parts of the input, is the key ingredient in the Transformer. Recent work has shown, however, that a large proportion of the heads in a Transformer’s multi-head attention mechanism can be safely pruned away without significantly harming the performance of the model; such pruning leads to models that are noticeably smaller and faster in practice. Our work introduces a new head pruning technique that we term differentiable subset pruning. ntuitively, our method learns per- head importance variables and then enforces a user-specified hard constraint on the number of unpruned heads. he importance variables are learned via stochastic gradient descent. e conduct experiments on natural language inference and machine translation; we show that differentiable subset pruning performs comparably or better than previous works while offering precise control of the sparsity level.1

Perbandingan Fluid Genetic Algorithm dan Genetic Algorithm untuk Penjadwalan Perkuliahan

The lecture schedule is a problem that belongs to the NP-Hard problem and multi-objective problem because it has several variables that affect the preparation of the schedule and has limitations that must be met. One solution that has been found is using a Genetic Algorithm (GA). GA has been proven to be able to provide a schedule that can meet limitations in scheduling. Besides, it also found a new concept of thought from GA, namely the Fluid Genetic Algorithm (FGA). The most visible difference between FGA and GA is that there is no mutation process in each iteration. FGA has a new stage, namely individual born and new constants, namely global learning rate, individual learning rate, and diversity rate. This concept of thinking was tested in previous studies and found that FGA is superior to GA for the problem of finding the optimum value of a predetermined function, but this function is not included in the multi-objective problem. In this study, the testing and comparison of FGA and GA were conducted for the problem of scheduling lectures at STMIK XYZ. Based on the results obtained, FGA can produce a schedule without any hard constraint violations. FGA can be used to solve multi-objective problems. FGA has a smaller number of generations than GA. However, overall GA is superior in producing schedules without any problems.

Penjadwalan Perkuliahan Dengan Metode Vertex Graph Coloring Dan Simulated Annealing

<p><em>In college, lecture scheduling is very important in lecturing process, because the activities of lecturers and students depend on lecture schedule. To solve the problem, use Vertex GraphColoring and Simulated Annealing. In Vertex Graph Coloring, look for neighboring and neighboring vertices. While on Simulated Annealing, look for space and exchange positions randomly. The merger of Vertex Graph Coloring and Simulated Annealing aims to create optimum lecture schedule by looking at hard constraint and soft constraint. Testing is done at Faculty of Pharmacy University of North Sumatra, by making schedule from manual to computerized, so it is expected to make the schedule optimally and able to avoid hardconstaint and soft constraint.</em></p>

Orebody Modeling from Non-Parallel Cross Sections with Geometry Constraints

In this paper, we present an improved approach to the surface reconstruction of orebody from sets of interpreted cross sections that allows for shape control with geometry constraints. The soft and hard constraint rules based on adaptive sampling are proposed. As only the internal and external position relations of sections are calculated, it is unnecessary to estimate the normal directions of sections. Our key contribution is proposing an iterative closest point correction algorithm. It can be used for iterative correction of the distance field based on the constraint rules and the internal and external position relations of the model. We develop a rich variety of geometry constraints to dynamically control the shape trend of orebody for structural geologists. As both of the processes of interpolation and iso-surface extraction are improved, the performance of this method is excellent. Combined with the interactive tools of constraint rules, our approach is shown to be effective on non-trivial sparse sections. We show the reconstruction results with real geological datasets and compare the method with the existing reconstruction methods.

Automatic velocity analysis using high-resolution hyperbolic Radon transform

Velocity analysis is crucial in reflection seismic data processing and imaging. Velocity picking is widely used in the industry for building the initial velocity model. When the size of the seismic data becomes extremely large, we cannot afford the corresponding human endeavor that is required by the velocity picking. In such situations, an automatic velocity-picking algorithm is highly demanded. We have developed a novel automatic velocity-analysis algorithm that is based on the high-resolution hyperbolic Radon transform. We formulate the automatic velocity-analysis problem as a constrained optimization problem. To solve the optimization problem with a hard constraint on the sparsity and distribution of the velocity spectrum, we relax it to a more familiar L1-regularized optimization problem in two steps. We use the iterative preconditioned least-squares method to solve the L1-regularized problem, and then we apply the hard constraint of the target optimization during the iterative inversion. Using synthetic and field-data examples, we determine the successful performance of our algorithm.