Top all-time · #2713
Mat Kelcey
Personal blog · matpalm.com/blogaidata
Mat Kelcey writes brain of mat kelcey..., a personal blog at matpalm.com about ai and data. Since 2009, 28 different Hacker News readers have linked it 33 times, most often “Counting Bees on a Raspberry Pi with a ConvNet”. It is still writing: the last post was on 2026-07-18, and you can follow it by feed.
Too few sampled comments (13) to say how often HN recommends it.?
Why HN links it
Linked on HN for technical posts on simhash and near-duplicate detection, plus experiments with Common Crawl data and a Wikipedia-links post. Commenters cite them as references for their own points.
Summarised by Claude from a sample of 13 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Most-linked posts?
- Counting Bees on a Raspberry Pi with a ConvNet4 people · 4 mentions · first linked 2018-05-18
- /resemblance/simhash/3 people · 4 mentions · first linked 2014-07-26
- Do all first links on Wikipedia lead to philosophy?3 people · 3 mentions · first linked 2011-08-21
- Cool Machine Learning Books2 people · 2 mentions · first linked 2020-11-02
- Tokenising the english text of 30TB common crawl2 people · 2 mentions · first linked 2011-12-12
- A brutally short intro to Theano word embeddings2 people · 2 mentions · first linked 2015-03-29
What Hacker News says?
Something not entirely quite unlike this http://matpalm.com/blog/2011/08/13/wikipedia-philosophy/
Comment linking /blog/2011/08/13/wikipedia-philosophy/
on “The 2016 Nobel Prize in Physics”Submitted Do all first links on Wikipedia lead to philosophy? as a story · 289 points
HN discussionIt does introduce new information. This is how semi-supervised learning works. Many words may be missed by the weak model, but can be inferred correctly from their context. Then you have new labels to train the weak model on, to make it better at those words. The way I'm describing it is probably not the optimal way to do it. I don't know if there is a better way. But the point is it must be possible to take advanta…
Comment linking /semi_supervised_naive_bayes/semi_supervised_bayes.html
on “Introducing the Open Images Dataset”Submitted Cool Machine Learning Books as a story · 248 points
HN discussionhttp://matpalm.com/resemblance/simhash/ Simhash algorithm, the LSH i knew about (which i mistakenly thought is LSH) works exactly like VW. It is ngrams + hash.
Comment linking /resemblance/simhash/
on “Reformer, the Efficient Transformer”Submitted Counting Bees on a Raspberry Pi with a ConvNet as a story · 177 points
HN discussionDownvoters, please see http://matpalm.com/resemblance/simhash/ https://en.wikipedia.org/wiki/SimHash Simhash, a type of local sensitive hashing - using hash functions on ngrammed data. That is exactly what Vowpal Wabbit does.
Comment linking /resemblance/simhash/
on “Reformer, the Efficient Transformer”Submitted Tokenising the english text of 30TB common crawl as a story · 72 points
HN discussionThis reminds me a lot of simhashing e.g. https://matpalm.com/resemblance/simhash/
Comment linking /resemblance/simhash/
on “Binary vector search is better than FP32 vectors”Submitted The Median of a Trillion Numbers as a story · 48 points
HN discussionThis blog post: http://matpalm.com/blog/2012/01/01/common_crawl_collocations... mentions that commoncrawl's data is last updated September 2010. Does anyone know if that's still the case?
Comment linking /blog/2012/01/01/common_crawl_collocations/
on “Common Crawl”Submitted A GAN to generate dithers minimising frame difference for a slow movie player as a story · 22 points
HN discussionThe passage quoted above refers to the way google plans to audit FLoC, not FLoC itself. Each browser is assigned a single floc ID, which is (for now) a 50-bit hash that is not human readable on its own. But the floc ID is intended to capture something meaningful about browsing behavior. Two users with similar browsing history are likely to have similar, if not the same, floc IDs. Floc IDs which are close together (h…
Comment linking /resemblance/simhash/
on “Google is testing its new ad targeting tech in millions of browsers”Submitted The median of a trillion numbers as a story · 11 points
HN discussionCool! @matpalm, please run your English extractor scripts on it! ( http://matpalm.com/blog/2011/12/10/common_crawl_visible_text... ) 'If you would like access to this set of crawl data, please contact us at info at archive dot org and let us know who you are and what you’re hoping to do with it. We may not be able to say “yes” to all requests, since we’re just figuring out whether this is a good idea, but everyone w…
Comment linking /blog/2011/12/10/common_crawl_visible_text/
on “80 terabytes of archived web crawl data available for research”Submitted A brutally short intro to Theano word embeddings as a story · 11 points
HN discussionThe author pointed this out; see "optimizing even more" under http://matpalm.com/median/erlang_multi.html
Comment linking /median/erlang_multi.html
on “The Median of a Trillion Numbers”Submitted Counting Bees on a Raspberry Pi with a ConvNet as a story · 3 points
HN discussionSource: http://matpalm.com/blog/counting_bees/ Once again, mods don't replace the link to an obvious blog spam article just because it's from TechCrunch.
Comment linking /blog/counting_bees/
on “Counting bees with a Raspberry Pi”Submitted Do all first links on Wikipedia lead to Philosophy? (2011) as a story · 2 points
HN discussionPrevious discussion: https://news.ycombinator.com/item?id=17103635 The blog post the article is actually about: http://matpalm.com/blog/counting_bees/
Comment linking /blog/counting_bees/
on “Counting bees with a Raspberry Pi”Submitted My updated list of cool machine learning books as a story · 2 points
HN discussionFor fast and larger scale near-duplicate detection see Google All-Pairs Similarity Search: https://code.google.com/p/google-all-pairs-similarity-search... . The SimHash algo: http://matpalm.com/resemblance/simhash/ and Gensim's Similarities Class: http://radimrehurek.com/gensim/similarities/docsim.html For blogspam detection can't PG whip up a nice Bayesian model for that ;) ? Again if you want fast and large scale…
Comment linking /resemblance/simhash/
on “Winnowing: Local Algorithms for Document Fingerprinting (2003) [pdf]”Submitted Learning to do laps with reinforcement learning and neural nets as a story · 2 points
HN discussionhave a look at Simhash as well: https://www.matpalm.com/resemblance/simhash/
Comment linking /resemblance/simhash/
on “MinHash – How to Find Similarity at Scale”Submitted Brutally short intro to theano word embeddings as a story · 2 points
HN discussionSubmitted Semi-supervised Learning with Naive Bayes as a story · 2 points
HN discussionSubmitted The median of a trillion numbers as a story · 2 points
HN discussionSubmitted Text classification - resemblance metrics as a story · 2 points
HN discussionSubmitted Collocations in Wikipedia - terms occurring frequently together as a story · 2 points
HN discussion
