Algorithms
Explicit colliding inputs for komihash, a5hash, MuseAir, t1ha2, HighwayHash and others under random seeds, measured as worst-case collision bits, next to hash functions with proven bounds and their speed.
Read moreAI for Hardware
Over 43 days, AI agents landed 2,968 commits on a fork of CIRCT: an event-driven Verilog simulator, UVM and cocotb support, bounded model checking, equivalence checking and mutation testing. Published on the Normal Computing blog.
Read moreProbability
A short, elementary route from n! = ∫ x^n e^{-x} to sharp Stirling bounds via a Gaussian change of variables—and a natural path to the full expansion.
Read moreStochastic Processes
This blog post delves into the Ornstein-Uhlenbeck (OU) process and its application in the analysis of linear systems. We explore how the OU process, when applied to a Positive Semi-Definite (PSD) matrix and a specific vector, leads to more efficient and accurate sampling strategies.
Read moreMachine Learning
Binary-search a scalar shift, reuse it in backprop via implicit differentiation, and you get a simple, exact top-k relaxation that plays nicely with PyTorch autograd.
Read moreLanguage Models
ZipLM uses Gzip to create a language model, however it is quite bad. This post shows how to use beam search to improve the output of ZipLM. When sampling sentences from "The Great Gatsby," the raw model produces near-gibberish. However, implementing beam search yields increasingly coherent sentences as the beam width increases, although performance is still poor compared to older models.
Read moreTensors
A technical guide on converting Einstein summation to tensor network operations.
Read moreGame AI
I’ve been meaning to learn about AIs for games like Poker or Liar’s Dice for a while. Recently while reading DeepMind’s article Player of Games, I thought I might be able to make something really simple that works well enough to be fun. “It shouldn’t take more than an afternoon,” I thought. Of course, I ended up spending much more time on it, but it did turn out to be fun and very simple.
Read moreData Structures
I recently published a preprint, together with Jakob Knudsen, describing a new, optimal data structure for set similarity search. An interesting way to look at this is as an Evolutionary Algorithm (aka genetic algorithm). We think it's probably the first time the provably best way to solve a problem has been evolutionary. This post won't go deep into the mathematics, but will contain code samples and some fun gadgets I built to simulate it.
Read moreProbability
This blog post sheds light on the less-discussed sampling without replacement scenario, its implications in algorithm analysis, and how it compares with its more famous counterpart, the binomial distribution.
Read moreNo posts match your search.
