Yann-Hervé Hellouvry’s Post

This paper proposes new schemes for #joint #channel and #data #estimation (#JCDE) and #radar #parameter #estimation (#RPE) in #doubly-#dispersive #channels, such that #integrated #sensing #and #communications (#ISAC) is enabled by #user #equipment (#UE) independently performing JCDE, and #base #stations (#BSs) performing RPE. The contributed JCDE and RPE schemes are designed for waveforms known to perform well in doubly-dispersive channels, under a unified model that captures the features of either legacy #orthogonal #frequency #division #multiplexing (#OFDM), #state-#of-#the-#art (#SotA) #orthogonal #time #frequency #space (#OTFS), and next-generation affine frequency division multiplexing (AFDM) systems. The proposed JCDE algorithm is based on a Bayesian #parametric #bilinear #Gaussian #belief #propagation (#PBiGaBP) framework first proposed for OTFS and here shown to apply to all aforementioned waveforms, while the RPE scheme is based on a new #probabilistic #data #association (#PDA) approach incorporating a Bernoulli-Gaussian denoising, optimized via #expectation #maximization (#EM). ----Kuranage Roche Rayan Ranasinghe, Hyeon Seok Rou， @Giuseppe Thadeu Freitas de Abreu, @Giuseppe Thadeu Freitas de Abreu, KENTA ITO More details can be found at this link: https://lnkd.in/gKeh_ejU

xLSTM achieves new SOTA for Remote Sensing Change Detection (identifying changes in objects of interest by comparing images of the same area taken at different times): https://lnkd.in/dgcPJBEK xLSTM "achieves state-of-the-art performance across three benchmark datasets, offering a compelling balance between efficiency and accuracy."

book of the day# 𝙈𝙚𝙖𝙨𝙪𝙧𝙚𝙢𝙚𝙣𝙩𝙨-𝘽𝙖𝙨𝙚𝙙 𝙍𝙖𝙙𝙖𝙧 𝙎𝙞𝙜𝙣𝙖𝙩𝙪𝙧𝙚 𝙈𝙤𝙙𝙚𝙡𝙞𝙣𝙜: 𝘼𝙣 𝘼𝙣𝙖𝙡𝙮𝙨𝙞𝙨 𝙁𝙧𝙖𝙢𝙚𝙬𝙤𝙧𝙠 (𝙈𝙄𝙏 𝙇𝙞𝙣𝙘𝙤𝙡𝙣 𝙇𝙖𝙗𝙤𝙧𝙖𝙩𝙤𝙧𝙮 𝙎𝙚𝙧𝙞𝙚𝙨) Characterizing objects (targets) from radar data and establishes a novel analysis framework for a class of signal processing techniques useful for high-resolution radar signature modeling. Comprehensive coverage related to basic radar concepts, signal representation, and radar measurements; the development of advanced analysis tools essential for high-resolution signature modeling; the development of novel wideband and narrowband radar imaging techniques; the application of 2D spectral estimation theory to wideband signal processing; ultra-wideband scattering phenomenology and sparse-band sensor data fusion; and the integration of field measurements into the radar signature modeling process. The analysis techniques developed in the text provide the framework for a novel approach, called measurements-based modeling (MBM), to model target signatures by incorporating measurement data into the signature model of the target. Extensive examples throughout compare the performance of the new techniques with that of conventional analysis techniques. https://lnkd.in/d5qTgpkT

🔍 With the Phased Array System Toolbox, we're not just simulating; we're matching real-world data with our models. Discover the power of accurate simulation in our latest exploration. #SystemToolbox #AccurateSimulation #TechInnovation

Do you think Brownian motion is the optimal driving noise process for your continuous-time diffusion model? We don’t think so! 💡We propose using Markov approximate fractional Brownian motion to improve image quality, pixel diversity, and distribution coverage compared to traditional, purely Brownian-driven diffusion models. I'm thrilled to announce that our paper "Generative Fractional Diffusion Models (GFDM)" has been accepted at #NeurIPS2024! 🚀 👉 Check-out our paper: https://lnkd.in/dqQavCMp 👉 Star our repository for code release: https://lnkd.in/d9z5KzKh 👉 Bookmark our poster session on Wednesday, 11 Dec 11 a.m. - 2 p.m. PST, in Vancouver: https://lnkd.in/dBb4SE-y The forward process of GFDM is driven by a Markov approximate fractional Brownian motion (MA-fBM) to control the "mild" or "wild" randomness in stochastic trajectories. MA-fBM provides control over long-term memory and roughness, allowing us to interpolate between the roughness of Brownian-driven SDEs and the underlying integration in PF ODEs, while also offering even rougher paths. 👇 By proposing augmented score matching, we show that learning a score model with the same input and output dimensions as the data is sufficient to approximate the score function of GFDM, allowing us to use the same model architecture as in traditional diffusion models. Our experiments demonstrate that, compared to purely Brownian dynamics, the super-diffusive (smooth) regime of MA-fBM yields higher image quality with fewer score model evaluations, improved pixel-wise diversity and better distribution coverage. I am grateful to have worked on this project with Maximilian Springenberg, Marco Aversa, Michael Detzel, Rembert Daems, Roderick Murray-Smith, Shinichi Nakajima, Sebastian Lapuschkin, Stefano Ermon, Tolga Birdal, Manfred Opper, Christoph Knochenhauer, Luis Oala, Wojciech Samek 🙏 It has been a true privilege to learn from such experts in computer science and mathematics throughout this journey 🙏 #NeurIPS #NeurIPS2024 #DiffusionModels #GenerativeModel

We are excited for our AI/ML team to present three (❗️) papers at NeurIPS this year! 🎉 One standout is a collaborative work titled 'Generative Fractional Diffusion Models (GFDM)' on improving generative image quality, pixel diversity, and distribution coverage, co-authored by our scientists Marco Aversa and Luis Oala with an exceptional global team, including Gabriel Nobis, Maximilian Springenberg, Michael Detzel, Rembert Daems, Roderick Murray-Smith, Shinichi Nakajima, Sebastian Lapuschkin, Stefano Ermon, Tolga Birdal, Manfred Opper, Christoph Knochenhauer, Luis Oala, Wojciech Samek This work brings together expertise from leading institutions Fraunhofer Heinrich Hertz Institute HHI, Dotphoton, Ghent University, University of Glasgow, Technische Universität Berlin, Stanford University, Imperial College London, Technical University of Munich. We're honoured and inspired to be part of this incredible community, advancing the future of imaging in AI. #NeurIPS2024 #DiffusionModels #AI #ML #GenAI #GenerativeAI

🔍 With the Phased Array System Toolbox, we're not just simulating; we're matching real-world data with our models. Discover the power of accurate simulation in our latest exploration. #SystemToolbox #AccurateSimulation #TechInnovation

🔍 With the Phased Array System Toolbox, we're not just simulating; we're matching real-world data with our models. Discover the power of accurate simulation in our latest exploration. #SystemToolbox #AccurateSimulation #TechInnovation

The Transitioning from static Building Information Modelling to dynamic Digital Twins 🏙️ This #DT paper is a treasure trove for anyone looking to overcome the lack of frameworks and protocols for sensor integration in Digital Twins. Credit to Andrea Revolti Pieter PauwelsPatrick Dallasega ———— Follow Me for #digitaltwins Links in My Profile Florian Huemer

🔍 With the Phased Array System Toolbox, we're not just simulating; we're matching real-world data with our models. Discover the power of accurate simulation in our latest exploration. #SystemToolbox #AccurateSimulation #TechInnovation