Browsing by Author "Cipiloglu Yildiz Z."

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    A portfolio construction framework using LSTM-based stock markets forecasting
    (John Wiley and Sons Ltd, 2022) Cipiloglu Yildiz Z.; Yildiz S.B.
    A novel framework that injects future return predictions into portfolio constructionstrategies is proposed in this study. First, a long–short-term-memory (LSTM) model is trained to learn the monthly closing prices of the stocks. Then these predictions are used in the calculation of portfolio weights. Five different portfolio construction strategies are introduced including modifications to smart-beta strategies. The suggested methods are compared to a number of baseline methods, using the stocks of BIST30 Turkey index. Our strategies yield a very high mean annualized return (25%) which is almost 50% higher than the baseline approaches. The mean Sharpe ratio of our strategies is 0.57, whereas the compared methods’ are 0.29 and −0.32. Comprehensive analysis of the results demonstrates that utilizing predicted returns in portfolio construction enables a significant improvement on the performance of the portfolios. © 2020 John Wiley & Sons Ltd.
  • No Thumbnail Available
    Item
    Learning a crowd-powered perceptual distance metric for facial blendshapes
    (Springer Science and Business Media Deutschland GmbH, 2023) Cipiloglu Yildiz Z.
    It is known that purely geometric distance metrics cannot reflect the human perception of facial expressions. A novel perceptually based distance metric designed for 3D facial blendshape models is proposed in this paper. To develop this metric, comparative evaluations of facial expressions were collected from a crowdsourcing experiment. Then, the weights of a distance metric, based on descriptive features of the models, were optimized to match the results with crowdsourced data, through a metric learning process. The method incorporates perceptual properties such as curvature and visual saliency. A formal analysis of the results proves the high correlation between the metric output and human perception. The effectiveness and success of the proposed metric were also compared to other distance alternatives. The proposed metric will enable intelligent processing of 3D facial blendshapes data in several ways. It will be possible to generate perceptually valid clustering and visualization of 3D facial blendshapes. It will help reduce storage and computational requirements by removing redundant expressions that are perceptually identical from the overall dataset. It can also be used to assist novice animators while creating plausible and expressive facial animations. © 2023, The Author(s).