Scopus İndeksli Yayınlar Koleksiyonu

Permanent URI for this collectionhttps://hdl.handle.net/20.500.12573/395

Browse

Filters

20243

Settings

Language: search.filters.language.enSubject: search.filters.subject.Deep Reinforcement Learning

Search Results

Now showing 1 - 3 of 3
  • Conference Object
    Citation - Scopus: 1
    From Traditional to Deep: Evaluating Sentiment Analysis Models on a Large-Scale Tweet Dataset
    (Institute of Electrical and Electronics Engineers Inc., 2024-10-26) Mammadov, Alisahib; Bakal, Gokhan
    This study investigates the effectiveness of various machine learning (ML) and deep learning (DL) techniques for large-scale sentiment analysis on Twitter data. We leverage a publicly available dataset of one million tweets, annotated with four sentiment labels (positive, negative, uncertainty, and liti-gious), to train and evaluate a range of models. Our experiments demonstrate that traditional ML algorithms, particularly XG-Boost, achieve high performance, with the best F1 score reaching 95.81% using a combination of unigrams and bigrams. Among DL models, a hybrid CNN-BiGRU architecture yields the highest average F1 score of 95.42%. Our findings highlight the strengths of different approaches for sentiment analysis on Twitter data and emphasize the importance of data preprocessing and model selection for achieving optimal performance. © 2025 Elsevier B.V., All rights reserved.
  • Conference Object
    Evaluating the Impact of Sentiment Analysis on Deep Reinforcement Learning-Based Trading Strategies
    (Institute of Electrical and Electronics Engineers Inc., 2024-10-26) Etcil, Mustafa; Kolukisa, Burak; Bakir-Güngör, Burcu
    Portfolio optimization is a form of investment management that aims to maximize returns while minimizing risks. However, the inherent complexity and unpredictability of financial markets pose a challenge. Recent advancements in machine learning, particularly in deep reinforcement learning (DRL), offer promising solutions by enabling dynamic and adaptive trading strategies. This paper presents a comprehensive evaluation of three actor-critic-based DRL algorithms-Advantage Actor-Critic (A2C), Deep Deterministic Policy Gradient (DDPG), and Proximal Policy Optimization (PPO)-applied to portfolio optimization. These strategies were implemented in both sentiment-aware and non-sentiment-aware versions, allowing for a direct comparison of their performance. The sentiment-aware models incorporated sentiment analysis using FinBERT and knowledge graphs to measure market sentiment from financial news, while the non-sentiment-aware models relied solely on stock prices and technical indicators. Our comparative study demonstrates that incorporating sentiment analysis resulted in consistently superior risk-adjusted returns and portfolio resilience during market fluctuations compared to non-sentiment-aware strategies. © 2025 Elsevier B.V., All rights reserved.
  • Article
    Citation - WoS: 21
    Citation - Scopus: 22
    ATOM: AI-Powered Sustainable Resource Management for Serverless Edge Computing Environments
    (IEEE-Inst Electrical Electronics Engineers Inc, 2024-11) Golec, Muhammed; Gill, Sukhpal Singh; Cuadrado, Felix; Parlikad, Ajith Kumar; Xu, Minxian; Wu, Huaming; Uhlig, Steve
    Serverless edge computing decreases unnecessary resource usage on end devices with limited processing power and storage capacity. Despite its benefits, serverless edge computing's zero scalability is the major source of the cold start delay, which is yet unsolved. This latency is unacceptable for time-sensitive Internet of Things (IoT) applications like autonomous cars. Most existing approaches need containers to idle and use extra computing resources. Edge devices have fewer resources than cloud-based systems, requiring new sustainable solutions. Therefore, we propose an AI-powered, sustainable resource management framework called ATOM for serverless edge computing. ATOM utilizes a deep reinforcement learning model to predict exactly when cold start latency will happen. We create a cold start dataset using a heart disease risk scenario and deploy using Google Cloud Functions. To demonstrate the superiority of ATOM, its performance is compared with two different baselines, which use the warm-start containers and a two-layer adaptive approach. The experimental results showed that although the ATOM required more calculation time of 118.76 seconds, it performed better in predicting cold start than baseline models with an RMSE ratio of 148.76. Additionally, the energy consumption and CO2 emission amount of these models are evaluated and compared for the training and prediction phases.