1D-CapsNet-LSTM: A Deep Learning-Based Model for Multi-Step Stock Index Forecasting DOI Creative Commons
Cheng Zhang, Nilam Nur Amir Sjarif, Roslina Ibrahim

et al.

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

Multi-step stock index forecasting is vital in finance for informed decision-making. Current methods on this task frequently produce unsatisfactory results due to the inherent data randomness and instability, thereby underscoring demand advanced models. Given superiority of capsule network (CapsNet) over CNN various classification tasks, study investigates potential integrating a 1D CapsNet with an LSTM multi-step forecasting. To end, hybrid 1D-CapsNet-LSTM model introduced, which utilizes generate high-level capsules from sequential capture temporal dependencies. maintain stochastic dependencies different horizons, multi-input multi-output (MIMO) strategy employed. The model's performance evaluated real-world market indices, including S&P 500, DJIA, IXIC, NYSE, compared baseline models, LSTM, RNN, CNN-LSTM, using metrics such as RMSE, MAE, MAPE, TIC. proposed consistently outperforms models two key aspects. It exhibits significant reductions errors Furthermore, it displays slower rate error increase lengthening forecast indicating increased robustness tasks.

Language: Английский

Pulmonary Nodule Detection, Segmentation and Classification Using Deep Learning: A Comprehensive Literature Review DOI Creative Commons

Ioannis Marinakis,

Konstantinos Κarampidis, Giorgos Papadourakis

et al.

BioMedInformatics, Journal Year: 2024, Volume and Issue: 4(3), P. 2043 - 2106

Published: Sept. 13, 2024

Lung cancer is a leading cause of cancer-related deaths worldwide, emphasizing the significance early detection. Computer-aided diagnostic systems have emerged as valuable tools for aiding radiologists in analysis medical images, particularly context lung screening. A typical pipeline diagnosis involves pulmonary nodule detection, segmentation, and classification. Although traditional machine learning methods been deployed previous years with great success, this literature review focuses on state-of-the-art deep methods. The objective to extract key insights methodologies from studies that exhibit high experimental results domain. This paper delves into databases utilized, preprocessing steps applied, data augmentation techniques employed, proposed exceptional outcomes. reviewed predominantly harness cutting-edge methodologies, encompassing convolutional neural networks (CNNs) advanced variants such 3D CNNs, alongside other innovative approaches Capsule transformers. examined these reflect continuous evolution datasets, discussed here collectively contribute development more efficient computer-aided systems, empowering dfhealthcare professionals fight against deadly disease.

Language: Английский

Citations

6
DeepFake video detection: Insights into model generalisation — A Systematic review DOI Creative Commons

Ramcharan Ramanaharan,

Deepani B. Guruge,

Johnson I. Agbinya

et al.

Data and Information Management, Journal Year: 2025, Volume and Issue: unknown, P. 100099 - 100099

Published: March 1, 2025

Language: Английский

Citations

0
A systematic literature review of video forgery detection techniques DOI

Manpreet Kaur Aulakh,

Navdeep Kanwal, Manish Bansal

et al.

Multimedia Tools and Applications, Journal Year: 2025, Volume and Issue: unknown

Published: April 2, 2025

Language: Английский

Citations

0
1D-CapsNet-LSTM: A deep learning-based model for multi-step stock index forecasting DOI Creative Commons
Cheng Zhang, Nilam Nur Amir Sjarif, Roslina Ibrahim

et al.

Journal of King Saud University - Computer and Information Sciences, Journal Year: 2024, Volume and Issue: 36(2), P. 101959 - 101959

Published: Feb. 1, 2024

Multi-step stock index forecasting is vital in finance for informed decision-making. Current methods this task frequently produce unsatisfactory results due to the inherent randomness and instability of data, thereby underscoring demand advanced models. Given superiority capsule network (CapsNet) over CNNs various classification tasks, study investigates potential integrating a 1D CapsNet with an LSTM multi-step forecasting. To end, hybrid 1D-CapsNet-LSTM model introduced, which utilizes generate high-level capsules from sequential data capture temporal dependencies. maintain stochastic dependencies different horizons, multi-input multi-output (MIMO) strategy employed. The model's performance evaluated on real-world market indices, including S&P 500, DJIA, IXIC, NYSE, compared baseline models, LSTM, RNN, CNN-LSTM, using metrics such as RMSE, MAE, MAPE, TIC. proposed consistently outperforms models two key aspects. It shows notable reductions errors when Additionally, it displays slower rate error escalation forecast horizons lengthen, suggesting enhanced robustness tasks.

Language: Английский

Citations

2
Manifesto of Deep Learning Architecture for Aspect Level Sentiment Analysis to extract customer criticism DOI Creative Commons
Nidhi Kushwaha, Bharat Singh, Sunil Agrawal

et al.

ICST Transactions on Scalable Information Systems, Journal Year: 2024, Volume and Issue: unknown

Published: April 9, 2024

Sentiment analysis, a critical task in natural language processing, aims to automatically identify and classify the sentiment expressed textual data. Aspect-level analysis focuses on determining at more granular level, targeting specific aspects or features within piece of text. In this paper, we explore various techniques for including traditional machine learning approaches state-of-the-art deep models. Additionally, has been utilized identifying extracting from text, addressing aspect-level ambiguity, capturing nuanced sentiments each aspect. These datasets are valuable conducting analysis. article, model based pre-trained neural networks. This can analyze sequences text as positive, negative, neutral without explicit human labeling. To evaluate these models, data Twitter's US airlines database was utilized. Experiments dataset reveal that BERT, RoBERTA DistilBERT outperforms than ML accuracy is efficient terms training time. Notably, our findings showcase significant advancements over previous methods rely supervised feature learning, bridging existing gaps methodologies. Our shed light challenges offering insights future research directions practical applications areas such customer feedback social media monitoring, opinion mining.

Language: Английский

Citations

1
Study on deviation correction and target positioning of intelligent operation robot for high-voltage switchgear DOI
Hao Wu, Nan Guo,

Chang Fan

et al.

Journal of the Brazilian Society of Mechanical Sciences and Engineering, Journal Year: 2024, Volume and Issue: 46(12)

Published: Nov. 4, 2024

Language: Английский

Citations

1
Distracted driver behavior recognition using modified capsule networks DOI Creative Commons

Jimmy Abdel Kadar,

Margareta Aprilia Kusuma Dewi,

Endang Suryawati

et al.

Mechatronics Electrical Power and Vehicular Technology, Journal Year: 2023, Volume and Issue: 14(2), P. 177 - 185

Published: Dec. 29, 2023

Human activity recognition (HAR) is an increasingly active study field within the computer vision community. In HAR, driver behavior can be detected to ensure safe travel. Detect behaviors using a capsule network with leave-one-subject-out validation. The was done CapsNet validation identify driving habits. proposed method in this consists of two parts, namely encoder and decoder. used modifies Sabour’s architecture by adding convolution layer before going primary layer. evaluated dataset 10 classes 300 images for each class. split based on hold-out resulting models were then compared conventional CNN architecture. objective research behavior. study, results accuracy rate 97.83 % However, decreased 53.11 when This because extracts all features including attributes participant contained input image (user-independent). Thus, model tends overfit.

Language: Английский

Citations

2
Hybrid Quantum–Classical Neural Networks for Efficient MNIST Binary Image Classification DOI Creative Commons

Deepak Ranga,

Sunil Prajapat, Zahid Akhtar

et al.

Mathematics, Journal Year: 2024, Volume and Issue: 12(23), P. 3684 - 3684

Published: Nov. 24, 2024

Image classification is a fundamental task in deep learning, and recent advances quantum computing have generated significant interest neural networks. Traditionally, Convolutional Neural Networks (CNNs) are employed to extract image features, while Multilayer Perceptrons (MLPs) handle decision making. However, parameterized circuits offer the potential capture complex features define sophisticated boundaries. In this paper, we present novel Hybrid Quantum–Classical Network (H-QNN) for classification, demonstrate its effectiveness using MNIST dataset. Our model combines with classical supervised learning enhance accuracy computational efficiency. study, detail architecture of H-QNN, emphasizing capability feature classification. Experimental results that proposed H-QNN outperforms conventional methods various training scenarios, showcasing high-dimensional tasks. Additionally, explore broader applicability hybrid quantum–classical approaches other domains. findings contribute growing body work machine underscore quantum-enhanced models recognition

Language: Английский

Citations

0
Iris Recognition via Deep Learning Using Capsule Networks with Enhanced Routing Algorithm DOI

Farzaneh Kuhifayegh,

Roozbeh Rajabi

Published: Dec. 1, 2024

Language: Английский

Citations

0
1D-CapsNet-LSTM: A Deep Learning-Based Model for Multi-Step Stock Index Forecasting DOI Creative Commons
Cheng Zhang, Nilam Nur Amir Sjarif, Roslina Ibrahim

et al.

arXiv (Cornell University), Journal Year: 2023, Volume and Issue: unknown

Published: Jan. 1, 2023

Multi-step stock index forecasting is vital in finance for informed decision-making. Current methods on this task frequently produce unsatisfactory results due to the inherent data randomness and instability, thereby underscoring demand advanced models. Given superiority of capsule network (CapsNet) over CNN various classification tasks, study investigates potential integrating a 1D CapsNet with an LSTM multi-step forecasting. To end, hybrid 1D-CapsNet-LSTM model introduced, which utilizes generate high-level capsules from sequential capture temporal dependencies. maintain stochastic dependencies different horizons, multi-input multi-output (MIMO) strategy employed. The model's performance evaluated real-world market indices, including S&P 500, DJIA, IXIC, NYSE, compared baseline models, LSTM, RNN, CNN-LSTM, using metrics such as RMSE, MAE, MAPE, TIC. proposed consistently outperforms models two key aspects. It exhibits significant reductions errors Furthermore, it displays slower rate error increase lengthening forecast indicating increased robustness tasks.

Language: Английский

Citations

0