Blog
LiveCC: Learning Video LLM with Streaming Speech Transcription at Scale
Recent video large language models (Video LLMs) often depend on costly human annotations or proprietary model APIs (per esempio., GPT-4o) to produce training data, which limits their training at scale. In questo documento, we explore large-scale training for Video LLM with cheap automatic...
LLMs meet Federated Learning for Scalable and Secure IoT Management
The rapid expansion of IoT ecosystems introduces severe challenges in scalability, security, and real-time decision-making. Traditional centralized architectures struggle with latency, privacy concerns, and excessive resource consumption, making them unsuitable for modern large-scale IoT deployments. This paper presents a novel Federated Learning-driven...
$η$-Einstein Sasakian Lie algebras
We study $\eta$-Einstein Sasakian structures on Lie algebras, questo è, Sasakian structures whose associated Ricci tensor satisfies an Einstein-like condition. We divide into the cases in which the Lie algebra’s centre is non-trivial (and necessarily one-dimensional) and those where it is zero....
LHCspin: a Polarized Gas Target for LHC
The goal of the LHCspin project is to develop innovative solutions for measuring the 3D structure of nucleons in high-energy polarized fixed-target collisions at LHC, exploring new processes and exploiting new probes in a unique, previously unexplored, kinematic regime. A precise multi-dimensional...
Rotational ultrasound and photoacoustic tomography of the human body
Imaging the human body’s morphological and angiographic information is essential for diagnosing, monitoring, and treating medical conditions. Ultrasonography performs the morphological assessment of the soft tissue based on acoustic impedance variations, whereas photoacoustic tomography (PAT) can visualize blood vessels based on intrinsic...
A Comparative and Measurement-Based Study on Real-Time Network KPI Extraction Methods for 5G and Beyond Applications
Key performance indicators (KPIs), which can be extracted from the standardized interfaces of network equipment defined by current standards, constitute a primary data source that can be leveraged in the development of non-standardized new equipment, architetture, and computational tools. In next-generation technologies,...
Linking the KM3-230213A Neutrino Event to Dark Matter Decay and Gravitational Waves Signals
The KM3NeT collaboration recently reported the detection of an ultra-high-energy neutrino event, dubbed KM3-230213A. This is the first observed neutrino event with energy of the order of $\mathcal{O}(100)$PeV, the origin of which remains unclear. We interprete this high energy neutrino event results...
Search for Axionlike Dark Matter Using Liquid-State Nuclear Magnetic Resonance
We search for dark matter in the form of axionlike particles (ALPs) in the mass range $5.576741 \,\matematica{neV/c^2}$ – $5.577733\,\matematica{neV/c^2}$ by probing their possible coupling to fermion spins through the ALP field gradient. This is achieved by performing proton nuclear magnetic resonance...
Certified Mitigation of Worst-Case LLM Copyright Infringement
The exposure of large language models (LLM) to copyrighted material during pre-training raises concerns about unintentional copyright infringement post deployment. This has driven the development of “copyright takedown” metodi, post-training approaches aimed at preventing models from generating content substantially similar to copyrighted...
PRIME: Fast Primal-Dual Feedback Optimization for Markets with Application to Optimal Power Flow
Online Feedback Optimization (OFO) controllers iteratively drive a plant to an optimal operating point that satisfies input and output constraints, relying solely on the input-output sensitivity as model information. This paper introduces PRIME (PRoximal Iterative MarkEts), a novel OFO approach based on...