Faculty of Natural Sciences and Mathematics

Permanent URI for this communityhttps://repository.ukim.mk/handle/20.500.12188/16

Browse

Settings

Author: search.filters.author.Kresoja, MilenaHas files: No

Search Results

Now showing 1 - 3 of 3
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Adaptive stochastic approximation algorithm
    (Springer Science and Business Media LLC, 2017-02-27)
    Kresoja, Milena
    ;
    Lužanin, Zorana
    ;
    In this paper, stochastic approximation (SA) algorithm with a new adaptive step size scheme is proposed. New adaptive step size scheme uses a fixed number of previous noisy function values to adjust steps at every iteration. The algorithm is formulated for a general descent direction and almost sure convergence is established. The case when negative gradient is chosen as a search direction is also considered. The algorithm is tested on a set of standard test problems. Numerical results show good performance and verify efficiency of the algorithm compared to some of existing algorithms with adaptive step sizes.
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Descent Direction Stochastic Approximation Algorithm with Adaptive Step Sizes
    (Global Science Press, 2019-06-01)
    Lužanin, Zorana
    ;
    ;
    Kresoja, Milena
    A stochastic approximation (SA) algorithm with new adaptive step sizes for solving unconstrained minimization problems in noisy environment is proposed. New adaptive step size scheme uses ordered statistics of fixed number of previous noisy function values as a criterion for accepting good and rejecting bad steps. The scheme allows the algorithm to move in bigger steps and avoid steps proportional to 1/k when it is expected that larger steps will improve the performance. An algorithm with the new adaptive scheme is defined for a general descent direction. The almost sure convergence is established. The performance of new algorithm is tested on a set of standard test problems and compared with relevant algorithms. Numerical results support theoretical expectations and verify efficiency of the algorithm regardless of chosen search direction and noise level. Numerical results on problems arising in machine learning are also presented. Linear regression problem is considered using real data set. The results suggest that the proposed algorithm shows promise.
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Increasing efficiency of on-line shopping by optimizing the staff schedule
    (Gran Sasso Science Institute, 2018-05-18)
    Bikov, Dusan
    ;
    Dvoriashyna, Mariia
    ;
    Ertugrul, Ümit
    ;
    Kresoja, Milena
    ;
    Koceva Lazarova, Limonka
    COOP Drive is an on-line shopping system recently started by COOP Liguria. Customers place their orders on-line, which are then processed by employees and collected at the time chosen by the customer. The problem proposed by COOP consists of two main parts: i) optimizing the staff schedule in COOP Drive ii) understanding if and to what extent such a schedule could be improved if orders were placed in advance. Providing a good schedule is very important for employees to reach an adequate level of satisfaction. According to the proposed problem from the on-line food shopping service, our aim was to make optimal staff scheduling such that each employee has `constant' working hours, i.e. that they work the same number of hours each working day. We introduce three different complementary models as approaches for the solution. The fi rst model is based on scheduling approach, which we solved for a simplifi ed scenario and is aimed to answer the first part of the problem. For the second question, we adopted two different approaches: an agent-based model that aims to understand the number of employees needed to process the order history and a worker placement model, that is developed to predict the number of employees required every hour to process the orders. These two models suggest that no signi ficant reduction of employees can be obtained by placing the orders in advance, however, signi ficant bene fit is achieved in terms of homogeneity of the schedule.