- Research Article
- 10.1007/s00034-026-03521-6
Multiclass Logistic Regression with Missing Gaussian Mixture Covariates
- Mar 06, 2026
- Circuits Systems and Signal Processing
- M Cherifi + 4 more +4
Publications from 2021 to 2026
Showing 10 of 1,688 papers
Multiclass Logistic Regression with Missing Gaussian Mixture Covariates
Convergence rate for the coupon collector’s problem with Stein’s method
An enhanced experimental paradigm for auditory BCIs by addressing both acoustic and human factors
This paper describes an experimental paradigm designed to evaluate the usability of non-sinusoidal amplitude-modulated stimuli for Brain-Computer Interfaces (BCIs) based on steady-state auditory evoked potentials. Instead of the commonly used pure tone, which can be uncomfortable, detectability of alternatives like Brownian noise or natural sounds (e.g. cicada song and cat’s purr), which are expected to be more pleasant to listen to, is assessed. Brain responses from 48 participants are recorded while listening to various amplitude-modulated sounds. To ensure accurate sound reproduction, the audio setup’s frequency response is measured and compensated for, using an Head And Torso Simulator (HATS). All stimuli are also equalized in loudness to prevent perceived level differences between stimuli. Participants’ hearing acuity is measured using an audiometric procedure developed for the needs of the experiment. Participants also completed questionnaires to collect information about their laterality, musicality, age and sex, as these factors may influence auditory perception.
Read moreReasons for malpractice claims after first revision of total hip arthroplasty in France: Insurance data from 263 consecutive claims from 2010 to 2023.
Creation and Comparison of NooJ Grammars for Temporal Expressions in Ukrainian and Serbian
Design and Characterization of Wide-Tuning CMOS Active Inductors for RF and Microwave Applications
This chapter presents the design, implementation, and evaluation of a tunable active inductor implemented in standard 130 nm CMOS technology. Two topologies are proposed: a single-ended implementation based on an enhanced cascode-grounded gyrator-C structure, and a differential topology incorporating negative-resistance compensation. In this design, independent inductance/quality factor (Q) tuning is achieved in the single-ended circuit through a dual control mechanism represented by the feedback resistor and bias currents, while the differential topology expands this concept to take advantage of cross-coupled devices for loss cancellation, offering enhanced Q and wide inductance tunability. The two structures were fabricated and measured, and both post-layout simulated and measured results exhibit compact silicon area, wide frequency tunability, excellent Q-factor performance, and low noise figures. The proposed tunable active inductors feature great robustness to PVT variations and are ideally suitable for integration into RF building blocks including filters, oscillators, and low-noise amplifiers.
Read moreSemantic Communication for Person Re-Identification
Wireless networks are undergoing a fundamental transformation from traditional data-centric architectures to intelligent service-based systems. This evolution is driven by advances in artificial intelligence that enable networks to support diverse applications such as human activity recognition, e-health services, and robotics. The current challenge lies in managing the massive data transmission requirements of these services. Thanks to deep learning models, networks can extract and transmit semantic representations instead of complete data sets, offering a more efficient approach to data transmission. However, these semantic representations face challenges when transmitted wireless, including interference, noise, and channel fading. This paper proposes to redesign information processing and transmission using a semantic-based approach for a person re-identification service. This includes studying communication errors on semantic accuracy, comparing encoding/decoding methods that account for network disruptions, and moving beyond the assumption of error-free semantic transmission.
Read moreThe Anatomy of Olympic Games: a Mobile Traffic Demand Perspective
Exploiting the Synergies of WLAN and Cellular Networks within the Wi-FIP Project
Wireless local area networks (WLAN) and recent Wi-Fi standards, are used since a few decades for a large variety of network applications. Since a few years, the $5^{\text {th }}$ generation wireless cellular networks (5 G) are being deployed and used, most of the time for similar applications. Both technologies and their evolution, are expected to be deployed within the wireless ecosystem by 2030 and for decades beyond. The Wi-FIP project invokes these two wireless networking solutions and their evolution, to further enhance the wireless data networks regarding sustainable and scalable quality of hybrid network services. This paper presents the visions and objectives of the Wi-FIP project, targeting to enhance synergies and integration of Wi-Fi in 5G/beyond 5G networks, through the convergent B2B/B2C usage model concept, sustainable multi-connectivity, hierarchical Software Defined Networking (SDN) and orchestration. The paper analyzes the functionalities to develop for sustainable, secure and hybrid WLAN/cellular networking.
Read moreNGT: A Retrieval-Augmented Language System for Telecoms Documentation Processing
NGT is a RAG framework specifically developed to address the challenges of applying large language models (LLMs) and recovered generation (RAG) systems in the telecommunication domain. It focuses on handling the complex nature of telecom standard documents, particularly 3rd generation Partnership Project (3GPP) documents. NGT aims to pave the way for the use of LLMs in telecommunications and provides guidelines for the implementation of RAGs in other fields. In this paper, the architecture of NGT is described, and a benchmarking review is presented, comparing the system to some reference Large Language Models (LLMs) and other systems.
Read more