The use of deep neural networks for traffic demand forecasting has garnered significant attention from both academic and industrial communities. Compared with the traditional traffic flow forecasting task, the Origin-...
详细信息
ISBN:
(纸本)9789819970186;9789819970193
The use of deep neural networks for traffic demand forecasting has garnered significant attention from both academic and industrial communities. Compared with the traditional traffic flow forecasting task, the Origin-Destination(OD) demand prediction task is more valuable and challenging, and several methods have been proposed for OD demand prediction. However, most existing methods follow a general technical route to aggregate historical information spatially and temporally. This paper proposes an alternative approach to predict Origin-Destination demand, named Zoom-based autoencoder for Origin-Destination demand prediction (ODZAE). The main objective of our research is to enhance the integration of diverse inherent patterns in real-world OD demand data in a more efficient manner. Besides, we proposed a zoom operation to learn spatial relationships between traffic nodes and 3DGCN to simultaneously model spatial and temporal dependencies. We have conducted experiments on two real-world datasets from Beijing Subway and New York Taxi, and the results demonstrate the superiority of our model against the state-of-art approaches.
Generally, the optimal shape depends on the characteristics of the magnetic materials used. In this study, we propose an efficient multi-objective topology optimization method that simultaneously considers material se...
详细信息
ISBN:
(纸本)9798350348958;9798350348965
Generally, the optimal shape depends on the characteristics of the magnetic materials used. In this study, we propose an efficient multi-objective topology optimization method that simultaneously considers material selection and shape change. The proposed approach utilizes an autoencoder, a type of deep learning model, to compress the information of both the device's shape and material characteristics into a latent space. The method enables us to efficiently derive superior combinations of shapes and materials for improving device's performances as pareto solutions. Some numerical examples demonstrate we can successfully carry out the practical device design by combining the proposed approach with the level set method.
This study utilizes the pre-processed fMRI data provided by the 2024 ICBHI challenge and excludes two other signal data types with missing information to construct a CNN model that can distinguish three emotion classe...
详细信息
ISBN:
(纸本)9783031863226;9783031863233
This study utilizes the pre-processed fMRI data provided by the 2024 ICBHI challenge and excludes two other signal data types with missing information to construct a CNN model that can distinguish three emotion classes and their corresponding levels. However, due to the high variability and noise in the pre-processed fMRI data, a simple CNN model alone cannot achieve good classification performance. This study addresses the issue of noisy data by proposing a two-stage deep learning model training framework. In the first stage, an autoencoder method is adopted, leveraging its ability to effectively encode and decode data to extract useful signal features from the noisy data for use in the subsequent second stage. In the second stage, the effective features obtained by the encoder are transferred, and the weights of the encoding layers are combined with a fully connected layer for model retraining. This study also analyzes different methods of transferring the weights of the encoding layers. The best model for this study achieved an error rate of only 0.3383 on the official evaluation metric.
In the evolving landscape of the Internet of Things (IoT), the susceptibility to cyber threats is widespread, emphasizing the critical need for robust security measures. This paper introduces an innovative anomaly det...
详细信息
ISBN:
(纸本)9798350309492;9798350309485
In the evolving landscape of the Internet of Things (IoT), the susceptibility to cyber threats is widespread, emphasizing the critical need for robust security measures. This paper introduces an innovative anomaly detection system based on a hybrid LSTM-autoencoder approach. Focused on protocol headers analysis in Packet Capture (PCAP) datasets, for robust anomaly detection, our model demonstrates high F1-score in anomalies detection with 99% and 96% on CICIDS2017 and on real network traffic, respectively. Refining our strategy, we address the intricacies of IoT environments, presenting a significant leap forward in intrusion detection for IoT networks.
This paper explains the main design decisions in the development of variational quantum time series models and denoising quantum time series autoencoders. Although we cover a specific type of quantum model, the proble...
详细信息
ISBN:
(纸本)9783031637773;9783031637780
This paper explains the main design decisions in the development of variational quantum time series models and denoising quantum time series autoencoders. Although we cover a specific type of quantum model, the problems and solutions are generally applicable to many other methods of time series analysis. The paper highlights the benefits and weaknesses of alternative approaches to designing a model, its data encoding and decoding, ansatz and its parameters, measurements and their interpretation, and quantum model optimization. Practical issues in training and execution of quantum time series models on simulators, including those that are CPU and GPU based, as well as their deployment on quantum machines, are also explored. All experimental results are evaluated, and the final recommendations are provided for the developers of quantum models focused on time series analysis.
Wireless federated learning (FL) relies on efficient uplink communications to aggregate model updates across distributed edge devices. Over-the-air computation (a.k.a. AirComp) has emerged as a promising approach for ...
详细信息
ISBN:
(纸本)9781728190549
Wireless federated learning (FL) relies on efficient uplink communications to aggregate model updates across distributed edge devices. Over-the-air computation (a.k.a. AirComp) has emerged as a promising approach for addressing the scalability challenge of FL over wireless links with limited communication resources. Unlike conventional methods, AirComp allows multiple edge devices to transmit uplink signals simultaneously, enabling the parameter server to directly decode the average global model. However, existing AirComp solutions are intrinsically analog, while modern wireless systems predominantly adopt digital modulations. Consequently, careful constellation designs are necessary to accurately decode the sum model updates without ambiguity. In this paper, we propose an end-to-end communication system supporting AirComp with digital modulation, aiming to overcome the challenges associated with accurate decoding of the sum signal with constellation designs. We leverage autoencoder network structures and explore the joint optimization of transmitter and receiver components. Our approach fills an important gap in the context of accurately decoding the sum signal in digital modulation-based AirComp, which can advance the deployment of FL in contemporary wireless systems.
Spatiotemporal prediction aims to generate future sequences by paradigms learned from historical contexts. It is essential in numerous domains, such as traffic flow prediction and weather forecasting. Recently, resear...
详细信息
ISBN:
(纸本)9798350344868;9798350344851
Spatiotemporal prediction aims to generate future sequences by paradigms learned from historical contexts. It is essential in numerous domains, such as traffic flow prediction and weather forecasting. Recently, research in this field has been predominantly driven by deep neural networks based on autoencoder architectures. However, existing methods commonly adopt autoencoder architectures with identical receptive field sizes. To address this issue, we propose an Asymmetric Receptive Field autoencoder (ARFA) model, which introduces corresponding sizes of receptive field modules tailored to the distinct functionalities of the encoder and decoder. In the encoder, we present a large kernel module for global spatiotemporal feature extraction. In the decoder, we develop a small kernel module for local spatiotemporal information reconstruction. Experimental results demonstrate that ARFA consistently achieves state-of-the-art performance on popular datasets. Additionally, we construct the RainBench, a large-scale radar echo dataset for precipitation prediction, to address the scarcity of meteorological data in the domain.
With recent text-to-image models, anyone can generate deceptively realistic images with arbitrary contents, fueling the growing threat of visual disinformation. A key enabler for generating high- resolution images wit...
详细信息
ISBN:
(纸本)9798350353006
With recent text-to-image models, anyone can generate deceptively realistic images with arbitrary contents, fueling the growing threat of visual disinformation. A key enabler for generating high- resolution images with low computational cost has been the development of latent diffusion models (LDMs). In contrast to conventional diffusion models, LDMs perform the denoising process in the low-dimensional latent space of a pre-trained autoencoder ( AE) instead of the high- dimensional image space. Despite their relevance, the forensic analysis of LDMs is still in its infancy. In this work we propose AEROBLADE, a novel detection method which exploits an inherent component of LDMs: the AE used to transform images between image and latent space. We find that generated images can be more accurately reconstructed by the AE than real images, allowing for a simple detection approach based on the reconstruction error. Most importantly, our method is easy to implement and does not require any training, yet nearly matches the performance of detectors that rely on extensive training. We empirically demonstrate that AEROBLADE is effective against state-of-the-art LDMs, including Stable Diffusion and Midjourney. Beyond detection, our approach allows for the qualitative analysis of images, which can be leveraged for identifying inpainted regions. We release our code and data at https://***/jonasricker/aeroblade.
The Brain Tumor (BT) is the formation of abnormal brain cells, few of which can develop into cancer. Early and prompt detection of disease and development of treatment strategies improve patients' wellbeing and li...
详细信息
ISBN:
(纸本)9798350391558;9798350379990
The Brain Tumor (BT) is the formation of abnormal brain cells, few of which can develop into cancer. Early and prompt detection of disease and development of treatment strategies improve patients' wellbeing and life span. Models are created using Deep Learning (DL) and Magnetic Resonance Imaging (MRI) to classify and diagnose brain tumours. This facilitates the effortless detection of brain tumours. The BT is usually caused through abnormal brain cell proliferation that damage the brain structure as well as eventually resulted in various stages of BT. Early BT discovery as well as timely treatment have reduced the mortality rate. Consideration of deep structure for analysis that have been used in both non-linear feature extraction and unsupervised learning, which relies significantly on the autoencoder (AE). This transfer learning is utilized to obtain better accuracy. Applications for the AE and its variations have been effective in many domains, including recommender systems, data production, pattern recognition, and computer vision. Moreover, the Convolutional AE (CAE) toolkits are addressed for better performance in detection of the brain tumor. The goal of this research is determining whether DL methods can be utilized to automate the detection process. This work uses Convolutional Neural Network (CNN) with VGG19 as the hybrid model to provide the better or maximum accuracy of 92.39% than the traditional Convolutional Neural Network (CNN) with VGG19. This study objective is to use TL with hybrid CNN techniques through this assessment and analysis to direct researchers and medical experts towards effective BT detection systems.
Investigating optical properties (OPs) is crucial in the field of biophotonics. Various techniques are available for deriving OPs, with inverse Monte Carlo simulations (IMCS) being the most advanced for ex-vivo contex...
详细信息
ISBN:
(纸本)9781510673380;9781510673397
Investigating optical properties (OPs) is crucial in the field of biophotonics. Various techniques are available for deriving OPs, with inverse Monte Carlo simulations (IMCS) being the most advanced for ex-vivo contexts. However, identifying the spectral behavior of each microscopic absorber and scatterer responsible for generating these OPs requires further experimentation. To tackle this issue, a customized autoencoder neural network (ANN) is suggested. The ANN computes OPs from measurements, where the bottleneck corresponds to the number of absorbers and scatterers. The presented ANN functions asymmetrically and computes the final OPs using a linear combination of absorbers and scatterers. Consequently, the decoder's weight corresponds to the constituent's OPs spectral behavior. Validation was conducted by utilizing intralipid as a scatterer and ink as an absorber. The employment of the decoder weights facilitated the successful extraction of the spectral shape of every constituent.
暂无评论