HNHacker News
TopNewBestAskShowJobs

cjauvin

1,981 karma · joined February 26, 2009

https://cjauvin.github.io

https://cjauvin.blogspot.com

submissionscomments
cjauvin··on Google Brain Residency Program – 7 months in and looking ahead
This might be the closest thing there is to it: https://www.udacity.com/course/deep-learning--ud730
cjauvin··on The 280-Year-Old Algorithm Inside Google Trips
A couple of years ago, I explored the Seven Bridges of Konigsberg problem with a set of interactive processing.js applets.. it might be of interest to some here to feel the impossibility of the task, before understanding it with the classic proof:

http://cjauvin.blogspot.ca/2012/02/implicit-bridges.html

cjauvin··on Seldomly Used SQL: The Bool Aggregate Functions
It's actually the first time I'm seeing the `group by 1, 2` syntax.. is that Postgres-specific?
cjauvin··on Have You Tried Using a Nearest Neighbor Search?
Have you looked at the simhashing/minhashing techniques for that?
cjauvin··on Deep Learning at x.ai
word2vec is an algorithm to produce meaningful "word embeddings", which is a vector representation in a usually high-dimensional space. t-SNE is a dimensionality-reduction algorithm. Both can be used together, as they serve different purposes.
cjauvin··on New Python REST API and CLI micro-framework
While browsing his other repos, I found this tool by the same author, which is also really great:

https://github.com/timothycrosley/isort

(Emacs integration works very well!)

cjauvin··on Records: Python library for making raw SQL queries to Postgres databases
I had planned to "show it to HN" soon, but the occasion is too good now: I have also created a thin "psycopg2 wrapper", pretty much in the same spirit as Records (but different in purpose), which allows you to perform SQL operations with plain data structures:

https://github.com/cjauvin/little_pger

I have been using this for myself for quite some time, but feedback would be very appreciated!

Edit: A cool thing about it (I believe) is that it provides an "upsert wrapper" which makes use of either the new PG 9.5 "on conclict" mechanism if it's available, or an "update if exists, insert if not" two-step fallback, for previous versions.

cjauvin··on Show HN: Simple and elegant Markdown-based resumes
The markdown-to-pdf resume toolchain I'm currently using is:

(1) Render with https://github.com/joeyespo/grip (with a few minor CSS tweaks)

(2) "Print" to PDF with Chrome

The documents it produces are quite sober, but I find it works very well.

cjauvin··on PostgreSQL 9.5: UPSERT, Row Level Security, and Big Data
I wonder how much time it will take to appear in the Ubuntu apt repo? Should it be already there (I don't see it yet)?

Edit: I meant apt.postgresql.org of course, not the official Ubuntu repo..

cjauvin··on Concepts for Your Cognitive Toolkit
Am I the only one not finding the connection between the Anthropic Principle and the Sleeping Beauty problem completely obvious?
cjauvin··on A Deep Learning Dissenter Thinks He Has a More Powerful AI Approach
I'm curious about that book (The Master Algorithm): I began reading it, but stopped early because I got the impression it would be too "entry-level" for me, thus a waste of my time. Should I consider keeping on with it, and why?
cjauvin··on Why Python 3 Exists
I know it's a deeply subjective and personal thing unlikely to help anyone other than me, but I've always successfully relied on the following mnemonic for this particular problem:

* unicode -> encode, u -> e (vowels)

* str -> decode, s -> d (consonants)

cjauvin··on Ask HN: What's your favorite online course?
Discrete Optimization https://www.coursera.org/course/optimization -- I really enjoyed this challenging class with a very dynamic teacher, and organized around a set of tough problems that you can tackle using a choice of optimization paradigms (e.g. you can decide to "specialize" in "local search" if you want, and try to solve eveything with it).
cjauvin··on Try out Stanford's CoreNLP natural language software
I'm curious to know what you mean by bAIb?
cjauvin··on Artificial General Intelligence via the CogPrime Architecture (2012) [pdf]
Probably this:

http://www.technologyreview.com/view/541276/deep-learning-ma...

http://arxiv.org/abs/1509.01549

cjauvin··on Building a Tetris Clone in X86 Assembly, Pt. Ⅰ: Hello, World
It's not 100% related but I had great fun building a very minimal Tetris clone in 6502 assembly, for the C=64, a couple of years ago (the code is "kinda" commented):

https://github.com/cjauvin/tetris-464

cjauvin··on Algorithm: The Hacker Movie
Thank you for the sunk cost fallacy reminder: because of it, I'm closing my browser tab after about 35 minutes. The premise is good and intriguing, but I agree with you that the movie is not very good, for the reasons you state.
cjauvin··on Ask HN: 16-hour work week jobs?
Thank you very much for this, I think it strikes a major chord with many in the "knowledge worker" community at large, and I hope it stays in the top spot for a while, and spawn a healthy discussion.

For my part I've been doing a mix of freelance and salaried work in the last few years (as a data scientist), and I have reached the conclusion that a strict 40-hour week is not an easy thing to handle. I think it stems from the fact that the "knowledge worker"/hacker ethos is often deeply rooted in a self-driven and motivated agenda of learning and experimenting with new things on a constant basis, very often just for the sake of it. This can easily conflict with a regular work schedule in terms of number of hours and commitment. But the point is that it really shouldn't, because very often, those two "modes" nourish each other, which can result in a stronger, more robust and ultimately more meaningful and happier work like.

I've been trying to convince my current boss that a 3-day week would be a more efficient and compressed use of my time, while giving me more time to pursue other contracts and projects. But it seems that there are some cultural barriers that makes it a difficult message to pass. I actually intend to use that thread to show that I'm not the only one in that situation, that such a culture really exists, and thus that it must be taken into account somehow.

It might be somewhat cliché, but I really believe that this mentality is a glimpse of the future in terms of work ethics.

cjauvin··on ExtJS: a dream or a nightmare? (2013)
I very much agree with all of your points. ExtJS is hard to beat in terms of the breadth of its widget set, layout configurations, data handling models, etc. Just for the almighty GridPanel, it's worth it I would say. But it's also more appropriate for internal, non-public facing apps. For that, you get pretty much all the usual benefits of a web application (ease of deployment, etc.) without getting lost in a traditional desktop GUI jungle. It's true that the library is frighteningly monolithic and the learning curve quite steep, but personally, I very much appreciate the declarative syntax, the very complete documentation, the multiple examples, the flexible MVC structure and the fact that it can easily result in nice decoupled code, as you can plug any backend you like, as long as it communicates through a well-defined protocol (I personally favor Python and Flask).
cjauvin··on A Tour of Machine Learning Algorithms
I wasn't aware of Association Rule Learning, which seems like a very interesting and useful concept.
cjauvin··on The utility of switching lanes when stuck in traffic
The philosopher Nick Bostrom has an interesting view about this question, based on the selection effect[1]: if you consider your presence in the traffic as a random sample from "observation acts", then you are more likely to be in the wrong lane (and hence would benefit from switching), simply because it is denser, i.e. it entails a higher probability mass for obversations of that type.

[1] http://plus.maths.org/content/cars-next-lane-really-do-go-fa...

cjauvin··on How the backpropagation algorithm works
This seems like it will be a very interesting book. If anyone is interested, I have written a short and compressed intro to NNs, using very simple Python code:

http://cjauvin.blogspot.ca/2013/10/neural-network-101.html

cjauvin··on Where should I start learning Assembly?
I concur that 6502 programming is great fun. When I had a bout of C=64 nostalgia recently, I created a simple Tetris clone as a relearning project:

https://github.com/cjauvin/tetris-c64

cjauvin··on Show HN: Ridiculously cheap bulk geocoding
I've been doing freelance geocoding gigs for a couple of institutions in the past years (canadian addresses), with only open source tools and data.

I also wrote a primer explaining the basic geocoding ideas:

http://cjauvin.blogspot.ca/2012/04/lean-geocoding.html

cjauvin··on Introduction to Artificial Neural Networks – Part 1
I have written a small series of posts which show (with some simple Python code and math) the relationship between the training procedure of an ANN (backpropagation) and some simpler, more basic machine learning algorithms (like logistic regression):

http://cjauvin.blogspot.ca/2013/10/neural-network-101.html

cjauvin··on Forget about sql in 50 lines of python code
I agree with all the comments saying that this is dangerous and not a good idea in general. However I've been doing (and using) something related (more in the spirit of Clojure HoneySQL, as I discovered recently), where the idea is to ease the interop of plain data structures and SQL in a pythonic way:

https://github.com/cjauvin/little_pger

I always got the nagging feeling that I should rather use SQLAlchemy, which probably solves the problem in a more elegant and general way, but this little thing has been working great for me.

cjauvin··on A magic square filled with consecutive primes
I recently posted a small example showing how to construct a magic square of odd degree using a surprisingly simple algorithm called De la Loubère (or sometimes also the Siamese method), with Python:

http://cjauvin.blogspot.ca/2013/09/siamese-dream.html

cjauvin··on Pony ORM - Use Pure Python to Speak to Your Data
I have solved a similar problem in a similar way (https://github.com/cjauvin/little_pger) by creating a very thin wrapper just above psycopg2, allowing to assemble SQL queries with kwargs and plain data structures.
cjauvin··on A sudoku solver in 33 lines of Clojure
Maybe this might help some helpless other Clojure newbees like me: I just compiled clojure and core.logic from the github repos, and it seems that the author just changed "everyo" to "everyg" very recently:

https://github.com/clojure/core.logic/commit/f831accb0659d04...

cjauvin··on Easy 6502 - Learn the 6502 Assembly Language
In fact, I don't know if what I did really corresponds to "real" double-buffering: at any point, the next move is created in an alternate buffer, which, when ready, is then switched as the visible portion of the video RAM (at vblank, using the raster interrupt). It works ok, but it's also quite possible that this mechanism might be a bit overkill.
← PreviousPage 6 of 7Next →