Tootfinder

Admissibility of Completely Randomized Trials: A Large-Deviation Approach
Guido Imbens, Chao Qin, Stefan Wager
https://arxiv.org/abs/2506.05329 https://

Admissibility of Completely Randomized Trials: A Large-Deviation Approach
When an experimenter has the option of running an adaptive trial, is it admissible to ignore this option and run a non-adaptive trial instead? We provide a negative answer to this question in the best-arm identification problem, where the experimenter aims to allocate measurement efforts judiciously to confidently deploy the most effective treatment arm. We find that, whenever there are at least three treatment arms, there exist simple adaptive designs that universally and strictly dominate non…

A late addition to stroke awareness month (May).
From my article: #health #stroke

#NowPlaying ❝This attention begs the question, why Japan and why noise? Perhaps the most simple answer is Japan is a very noisy place indeed.❞
From “Oh! Moro Volume 5” (1992) Kansai new art video magazine with scene report and interviews by David Hopkins/Public Bath.

This https://arxiv.org/abs/2503.01779 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_mat…

Curvature, macroscopic dimensions, and symmetric products of surfaces
We present a detailed study of the curvature and symplectic asphericity properties of symmetric products of surfaces. We show that these spaces can be used to answer nuanced questions arising in the study of closed Riemannian manifolds with positive scalar curvature. For example, we prove that symmetric products of surfaces sharply distinguish between two distinct notions of macroscopic dimension introduced by Gromov and the second-named author. As a natural generalization of this circle of ide…

Moment Sampling in Video LLMs for Long-Form Video QA
Mustafa Chasmai, Gauri Jagatap, Gouthaman KV, Grant Van Horn, Subhransu Maji, Andrea Fanelli
https://arxiv.org/abs/2507.00033 …

Moment Sampling in Video LLMs for Long-Form Video QA
Recent advancements in video large language models (Video LLMs) have significantly advanced the field of video question answering (VideoQA). While existing methods perform well on short videos, they often struggle with long-range reasoning in longer videos. To scale Video LLMs for longer video content, frame sub-sampling (selecting frames at regular intervals) is commonly used. However, this approach is suboptimal, often leading to the loss of crucial frames or the inclusion of redundant inform…

Cascadia: A Cascade Serving System for Large Language Models
Youhe Jiang, Fangcheng Fu, Wanru Zhao, Stephan Rabanser, Nicholas D. Lane, Binhang Yuan
https://arxiv.org/abs/2506.04203

Cascadia: A Cascade Serving System for Large Language Models
Recent advances in large language models (LLMs) have intensified the need to deliver both rapid responses and high-quality answers. More powerful models yield better results but incur higher inference latency, whereas smaller models are faster yet less capable. Recent work proposes balancing this latency-quality trade-off using model cascades, which route simpler queries to smaller models and more complex ones to larger models. However, enabling efficient cascade serving remains challenging. Cu…

Attention Is Not Always the Answer: Optimizing Voice Activity Detection with Simple Feature Fusion
Kumud Tripathi, Chowdam Venkata Kumar, Pankaj Wasnik
https://arxiv.org/abs/2506.01365

Attention Is Not Always the Answer: Optimizing Voice Activity Detection with Simple Feature Fusion
Voice Activity Detection (VAD) plays a key role in speech processing, often utilizing hand-crafted or neural features. This study examines the effectiveness of Mel-Frequency Cepstral Coefficients (MFCCs) and pre-trained model (PTM) features, including wav2vec 2.0, HuBERT, WavLM, UniSpeech, MMS, and Whisper. We propose FusionVAD, a unified framework that combines both feature types using three fusion strategies: concatenation, addition, and cross-attention (CA). Experimental results reveal that …

Beyond Sociodemographic Prompting: Using Supervision to Align LLMs with Human Response Distributions
Gauri Kambhatla, Sanjana Gautam, Angela Zhang, Alex Liu, Ravi Srinivasan, Junyi Jessy Li, Matthew Lease
https://arxiv.org/abs/2507.00439

Beyond Sociodemographic Prompting: Using Supervision to Align LLMs with Human Response Distributions
The ability to accurately predict how different population groups would answer subjective questions would have great value. In this work, we show that use of relatively simple supervision can greatly improve language model alignment with diverse population groups, as measured over three datasets spanning various topics. Beyond evaluating average performance, we also report how alignment varies across specific groups. The simplicity and generality of our approach promotes easy adoption, while ou…

This https://arxiv.org/abs/2003.11202 has been replaced.
link: https://scholar.google.com/scholar?q=a

Set System Blowups
We prove that given a constant $k \ge 2$ and a large set system $\mathcal{F}$ of sets of size at most $w$, a typical $k$-tuple of sets $(S_1, \cdots, S_k)$ from $\mathcal{F}$ can be ``blown up" in the following sense: for each $1 \le i \le k$, we can find a large subfamily $\mathcal{F}_i$ containing $S_i$ so that for $i \neq j$, if $T_i \in \mathcal{F}_i$ and $T_j \in \mathcal{F}_j$ , then $T_i \cap T_j=S_i \cap S_j$. We also show that the answer to the multicolor version of the sunflower conje…

This https://arxiv.org/abs/2405.17286 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_mat…

Asymptotics of extensions of simple $\mathbb Q$-algebras
We answer various questions concerning the distribution of extensions of a given central simple algebra $K$ over a number field. Specifically, we give asymptotics for the count of inner Galois extensions $L/K$ of fixed degree and center with bounded discriminant. We also relate the distribution of outer extensions of $K$ to the distribution of field extensions of its center $Z(K)$. This paper generalizes the study of asymptotics of field extensions to the noncommutative case in an analogous man…