No abstract available.
Front Matter
Front Matter
Energy Complexity of Fully-Connected Layers
The energy efficiency of processing convolutional neural networks (CNNs) is crucial for their deployment on low-power mobile devices. In our previous work, a simplified theoretical hardware-independent model of energy complexity for CNNs has been ...
Scalable Convolutional Neural Networks for Decoding of Terminated Convolutional Codes
We present a convolutional neural network (CNN) for the decoding of a terminated convolutional code (CC). For this use cases, an unlimited amount of labeled training data can be generated. However, the number of code words, i.e., pattern, to be ...
Optimizing an IDS (Intrusion Detection System) by Means of Advanced Metaheuristics
Intrusion Detection Systems (IDSs) are a primary research area in Cybersecurity nowadays. These are programs or methods designed to monitor and analyze network traffic aiming to identify suspicious patterns/attacks. MSNM (Multivariate Statistical ...
Iterative Graph Embedding and Clustering
Graph embedding can be seen as a transformation of any graph into low-dimensional vector space, where each vertex of the graph has a one-to-one correspondence with a vector in that space. The latest study in this field shows a particular interest ...
Boosting NSGA-II-Based Wrappers Speedup for High-Dimensional Data: Application to EEG Classification
The considerable technological evolution during the last deca-des has made it possible to deal with biological datasets of increasing higher dimensionality, such as those used in BCI applications. Thus, techniques such as feature selection, which ...
Extending Drift Detection Methods to Identify When Exactly the Change Happened
Data changing, or drifting, over time is a major problem when using classical machine learning on data streams. One approach to deal with this is to detect changes and react accordingly, for example by retraining the model. Most existing drift ...
Pedestrian Multi-object Tracking Algorithm Based on Attention Feature Fusion
Multi-Object Tracking (MOT) is a challenging research area in computer vision with significant practical applications. With the advent of deep neural networks, significant progress has been made in MOT, and Qdtrack has become a widely used ...
Fairness-Enhancing Ensemble Classification in Water Distribution Networks
As relevant examples such as the future criminal detection software [1] show, fairness of AI-based and social domain affecting decision support tools constitutes an important area of research. In this contribution, we investigate the applications ...
Front Matter
Observe Locally, Classify Globally: Using GNNs to Identify Sparse Matrix Structure
The performance of sparse matrix computation highly depends on the matching of the matrix format with the underlying structure of the data being computed on. Different sparse matrix formats are suitable for different structures of data. Therefore, ...
Unsupervised Topic Modeling with BERTopic for Coarse and Fine-Grained News Classification
Transformer models have achieved state-of-the-art results for news classification tasks, but remain difficult to modify to yield the desired class probabilities in a multi-class setting. Using a neural topic model to create dense topic clusters ...
Double Transfer Learning to Detect Lithium-Ion Batteries on X-Ray Images
With the soaring popularity of electronic gadgets, Lithium-Ion Batteries (LIB) have witnessed a remarkable surge. The inspiration behind this study arises from the urgent need to automate the identification of batteries in diverse contexts, such ...
MLFEN: Multi-scale Long-Distance Feature Extraction Network
Hyperspectral image fusion frequently leverages panchromatic and multispectral images. Although remote sensing images exhibit multi-scale features, prior research has predominantly focused on local feature extraction using convolutional approaches,...
Comparison of Fourier Bases and Asymmetric Network Bases in the Bio-Inspired Networks
Machine learning, deep learning and neural networks are extensively developed in many fields, in which neural network architectures have shown a variety of applications. However, there is a need for explainable fundamentals in complex neural ...
Temporal Dynamics of Drowsiness Detection Using LSTM-Based Models
Different LSTM-based models were tested for binary drowsiness detection using the ULg Multimodality Drowsiness Database (DROZY). The dataset contains physiological signals and behavioral measures collected from participants during different ...
Fine-Tuned SqueezeNet Lightweight Model for Classifying Surface Defects in Hot-Rolled Steel
- Francisco López de la Rosa,
- José Luis Gómez-Sirvent,
- Lidia María Belmonte,
- Rafael Morales,
- Antonio Fernández-Caballero
The advent of powerful artificial intelligence-based tools is opening up new opportunities to improve the efficiency of processes in the manufacturing industry. One of those processes is visual inspection, where deep learning approaches, ...
Shot Boundary Detection with Augmented Annotations
In recent years, deep learning approaches have been considered to provide state-of-the-art results in shot boundary detection. These approaches revolve around the need for large annotated datasets. The quality of the annotations is crucial to the ...
3D Human Body Models: Parametric and Generative Methods Review
This paper provides an overview of the current state-of-the-art in the field of 3D human body model estimation, reconstruction, and generation in computer vision. The paper focuses on the most widely used parametric and generative methods and ...
Deep Learning Recommendation System for Stock Market Investments
The paper proposes and compares two models for creating a recommendation system in the stock market, based on convolutional neural networks (CNN). The first model encodes the values of the time series of the stock exchange quotations into multiple ...
Minimal Optimal Region Generation for Enhanced Object Detection in Aerial Images Using Super-Resolution and Convolutional Neural Networks
In recent years, object detection has experienced impressive progress in several fields. However, identifying objects in aerial images remains a complex undertaking due to specific challenges, including the presence of small objects or tightly ...
Long-Term Hail Risk Assessment with Deep Neural Networks
Hail risk assessment is crucial for businesses, particularly in the agricultural and insurance sectors, as it helps estimate and mitigate potential losses. Although significant attention has been given to short-term hail forecasting, the lack of ...
Video Scene Segmentation Based on Triplet Loss Ranking
Scene segmentation is the task of segmenting the video in groups of frames with a high degree of semantic similarity. In this paper, we contribute to the task of video scene segmentation with the creation of a novel dataset for temporal scene ...
A Model for Classifying Emergency Events Based on Social Media Multimodal Data
Social media has emerged as a crucial source of information for emergency management. However, the diverse range of data types, including textual and visual information, presents a significant challenge for scholars seeking to analyze this ...
A Performance Evaluation of Lightweight Deep Learning Approaches for Bird Recognition
- Dmitrij Teterja,
- Jose Garcia-Rodriguez,
- Jorge Azorin-Lopez,
- Esther Sebastian-Gonzalez,
- Srdjan Krco,
- Dejan Drajic,
- Dejan Vukobratovic
Reliable identification of bird species is a critical task for many applications, such as conservation biology, biodiversity assessments, and monitoring bird populations. However, identifying birds in the wild by visual observation can be time-...