Blogue
MOOSComp: Improving Lightweight Long-Context Compressor via Mitigating Over-Smoothing and Incorporating Outlier Scores
Recent advances in large language models have significantly improved their ability to process long-context input, but practical applications are challenged by increased inference time and resource consumption, particularly in resource-constrained environments. Pour relever ces défis, we propose MOOSComp, a token-classification-based long-context compression...
Toroidal black holes in four dimensions
From a purely geometric (kinematic) perspective, black holes in four dimensional spacetimes can have event horizons with arbitrary topologies. It is only when energy conditions are imposed that the horizon’s shape is constrained to be spherical. Despite this, exploring exotic horizon topologies...
Radiometer Calibration using Machine Learning
Radiometers are crucial instruments in radio astronomy, forming the primary component of nearly all radio telescopes. They measure the intensity of electromagnetic radiation, converting this radiation into electrical signals. A radiometer’s primary components are an antenna and a Low Noise Amplifier (LNA),...
Preemption Aware Task Scheduling for Priority and Deadline Constrained DNN Inference Task Offloading in Homogeneous Mobile-Edge Networks
This paper addresses the computational offloading of Deep Neural Networks (DNNs) to nearby devices with similar processing capabilities, to avoid the larger communication delays incurred for cloud offloading. We present a preemption aware scheduling approach for priority and deadline constrained task offloading...
A self-avoiding curve associated with sums of digits
For each $n\in N ^{\ast }$, we write $s_{n}=\left( 1,\ldots ,1,0\droite) $ with $n$ times $1$. For each $a \in N$, we consider the binary representation $\left( a_{je}\droite) _{i\in -N }$ of $a$ with $a_{je}=0$ for nearly each $i$; we denote by...
Unsupervised Time-Series Signal Analysis with Autoencoders and Vision Transformers: A Review of Architectures and Applications
The rapid growth of unlabeled time-series data in domains such as wireless communications, radar, biomedical engineering, and the Internet of Things (IdO) has driven advancements in unsupervised learning. This review synthesizes recent progress in applying autoencoders and vision transformers for unsupervised signal...
Coalescence delay mediated by the gas layer during the impact of hot droplets
Coalescence may not occur immediately when droplets impact a liquid film. Despite the prevalence of the high-temperature condition during the impact process in many applications, the effect of droplet temperature on droplet coalescence is rarely considered. Dans cette étude, we experimentally investigate...
Cyclic Zoom: Multi-scale GRMHD Modeling of Black Hole Accretion and Feedback
We present a “cyclic zoom” method to capture the dynamics of accretion flows onto black holes across a vast range of spatial and temporal scales in general relativistic magnetohydrodynamic (GRMHD) simulation. In this method, we cyclically zoom out (derefine) and zoom in...
Graph modification of bounded size to minor-closed classes as fast as vertex deletion
A replacement action is a function $\mathcal{L}$ that maps each graph $H$ to a collection of graphs of size at most $|V(H)|$. Given a graph class $\mathcal{H}$, we consider a general family of graph modification problems, called $\mathcal{L}$-Replacement to $\mathcal{H}$, where the...
Constructing Four-Body Ballistic Lunar Transfers via Analytical Energy Conditions
This paper derives and summarizes the analytical conditions for lunar ballistic capture and constructs ballistic lunar transfers based on these conditions. We adopt the Sun-Earth/Moon planar bicircular restricted four-body problem as the dynamical model to construct lunar transfers. D'abord, the analytical conditions...