Tootfinder

Filipino workers in Taiwan's chip industry report struggling with overwork and discrimination, citing overnight shifts of up to 16 hours and abusive treatment (Rest of World)
https://restofworld.org/2025/filipino-workers-taiwan-chip-industry/

Workers describe abusive treatment in Taiwan’s semiconductor factories
A record number of people are leaving the Philippines to work in Taiwanese factories, where they face long shifts, low pay, and unequal treatment.

Reality of raising rabbits for meat in Spain. We reveal what the livestock industry doesn't want you to know. Now it's your turn to change this reality, sign the petition! #AnimalRights

Let's shut down factory farms once and for all
Industrial farming is devouring the planet and is the cruelest and dirtiest face of the business that profits from the lives of animals. These huge meat factories are responsible for climate change, unsustainability and cruelty to animals. With your signature, you help us to put an end to them.

System-Aware Unlearning Algorithms: Use Lesser, Forget Faster
Linda Lu, Ayush Sekhari, Karthik Sridharan
https://arxiv.org/abs/2506.06073 https://

System-Aware Unlearning Algorithms: Use Lesser, Forget Faster
Machine unlearning addresses the problem of updating a machine learning model/system trained on a dataset $S$ so that the influence of a set of deletion requests $U \subseteq S$ on the unlearned model is minimized. The gold standard definition of unlearning demands that the updated model, after deletion, be nearly identical to the model obtained by retraining. This definition is designed for a worst-case attacker (one who can recover not only the unlearned model but also the remaining data samp…

https://www.theverge.com/news/680893/nintendo-switch-2-eshop-faster look, I just want to know what the tech stack of the old shop was 😅

Writing a unit test for a newer, faster version of a libscopehal primitive and it wasn't lining up with the original results.
Turns out there was a bug in the original (FindZeroCrossings would fail to correctly detect a crossing between samples 0 and 1).

FERRET: Private Deep Learning Faster And Better Than DPSGD
David Zagardo
https://arxiv.org/abs/2506.05416 https://arxiv.org/pdf/2506.…

FERRET: Private Deep Learning Faster And Better Than DPSGD
We revisit 1-bit gradient compression through the lens of mutual-information differential privacy (MI-DP). Building on signSGD, we propose FERRET--Fast and Effective Restricted Release for Ethical Training--which transmits at most one sign bit per parameter group with Bernoulli masking. Theory: We prove each fired group leaks at most ln 2 nats; after subsampling with rate s, the total privacy loss of G groups trained for T steps with firing probability p is epsilon = G * T * s * p * ln 2. Thu…

The Spurious Factor Dilemma: Robust Inference in Heavy-Tailed Elliptical Factor Models
Jiang Hu, Jiahui Xie, Yangchun Zhang, Wang Zhou
https://arxiv.org/abs/2506.05116

The Spurious Factor Dilemma: Robust Inference in Heavy-Tailed Elliptical Factor Models
Factor models are essential tools for analyzing high-dimensional data, particularly in economics and finance. However, standard methods for determining the number of factors often overestimate the true number when data exhibit heavy-tailed randomness, misinterpreting noise-induced outliers as genuine factors. This paper addresses this challenge within the framework of Elliptical Factor Models (EFM), which accommodate both heavy tails and potential non-linear dependencies common in real-world da…

Un peu de musique pour égayer la journée 🎶:
https://suno.com/song/f936fa54-3c52-42b9-916a-eedbfe377b99?sh=FEJT9Wr8fUoaD77w
[Breakdown]
Aime-moi, déteste-moi—c’est pareil
Je vis dans ta tête, sans payer de loyer
Peu importe combien tu essaies de me repousser
Je ne fais que gagner
Parce que chaque mot que tu craches devient mon champagne

At a clandestine math conclave in Berkeley in May, a chatbot powered by o4-mini answered some of the hardest solvable problems much faster than a mathematician (Lyndie Chiou/Scientific American)
https://www.scientificamerican.com/article…

Inside the Secret Meeting Where Mathematicians Struggled to Outsmart AI
The world's leading mathematicians were stunned by how adept artificial intelligence is at doing their jobs

Finished the initial GPU based zero-crossing search block for libscopehal.
It's not used in anything yet as it's not an exact drop-in replacement API wise for the CPU version, but is massively faster (30x speedup for 50M points on 2080 Ti vs Xeon 6144) so I'll be gradually refactoring filters to take advantage of it,