Hanskunatai, Anantaporn

Loading...
Thumbnail Image
Preferred name
Hanskunatai, Anantaporn
Alternative Name
Hanskunatai, Anantapom
Main Affiliation
Email
anantaporn.ha@kmitl.ac.th

Settings

Now showing 1 - 2 of 2
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Hierarchical Convolutional Neural Networks using CCP-3 Block Architecture for Apparel Image Classification
    (2023-01-01)
    Chamnong, Natthamon
    ;
    Werapun, Jeeraporn
    ;
    Hanskunatai, Anantaporn
    In fashion applications, deep learning has been applied automatically to recognize and classify the apparel images under the massive visual data, emerged on social networks. To classify the apparel correctly and quickly is challenging due to a variety of apparel features and complexity of the classification. Recently, the hierarchical convolutional neural networks (H–CNN) with the VGGNet architecture was proposed to classify the fashion-MNIST datasets. However, the VGGNet (many layers) required many filters (in the convolution layer) and many neurons (in the fully connected layer), leading to computational complexity and long training-time. Therefore, this paper proposes to classify the apparel images by the H–CNN in cooperated with the new shallow-layer CCP-3-Block architecture, where each building block consists of two convolutional layers (CC) and one pooling layer (P). In the CCP-3-Block, the number of layers can be reduced (in the network), the number of filters (in the convolution layer), and the number of neurons (in the fully connected layer), while adding a new connection between the convolution layer and the pooling layer plus a batch-normalization technique before passing the activation so that networks can learn independently and train quickly. Moreover, dropout techniques were utilized in the feature mapping and fully connected to reduce overfitting, and the optimizer adaptive moment estimation was utilized to solve the decaying of gradients, which can improve the network-performance. The experimental results showed that the improved H–CNN model with our CCP-3-Block outperformed the recent H–CNN model with the VGGNet in terms of decreased loss, increased accuracy, and faster training.
  • Loading...
    Thumbnail Image
    Some of the metrics are blocked by your 
    consent settings
    Item type:Publication,
    Minimum cost of job assignment in polynomial time by adaptive unbiased filtering and branch-and-bound algorithm with the best predictor
    (2025-06-01)
    Werapun, Jeeraporn
    ;
    Towongpaichayont, Witchaya
    ;
    Hanskunatai, Anantaporn
    The minimum cost of job assignment (Min-JA) is one of the practical NP-hard problems to manage the optimization in science-and-engineering applications. Formally, the optimal solution of the Min-JA can be computed by the branch-and-bound (BnB) algorithm (with the efficient predictor) in O(n!), n = problem size, and O(n<sup>3</sup>) in the best case but that best case hardly occurs. Currently, metaheuristic algorithms, such as genetic algorithms (GA) and swarm-optimization algorithms, are extensively studied, for polynomial-time solutions. Recently, unbiased filtering (in search-space reduction) could solve some NP-hard problems, such as 0/1-knapsack and multiple 0/1-knapsacks with Latin square (LS) of m-capacity ranking, for the ideal solutions in polynomial time. To solve the Min-JA problem, we propose the adaptive unbiased-filtering (AU-filtering) in O(n<sup>3</sup>) with a new hybrid (search-space) reduction (of the indirect metaheuristic strategy and the exact BnB). Innovation-and-contribution of our AU-filtering is achieved through three main steps: 1. find 9 + n effective job-orders for the good initial solutions (by the indirect assignment with UP: unbiased predictor), 2. improve top 9-solutions by the indirect improvement of the significant job-orders (by Latin square of n permutations plus n complex mod-functions), and 3. classify objects (from three of the best solutions) for AU-filtering (on large n) with deep-reduction (on smaller n’) and repeat (1)-(3) until n’ < 6, the exact BnB is applied. In experiments, the proposed AU-filtering was evaluated by a simulation study, where its ideal results outperformed the best results of the hybrid swarm-GA algorithm on a variety of 2D datasets (n ≤ 1000).