Fragoel2

If only managers would perceive the money spent to pay such organizations as a necessity rather than burned cash

Suppose I would want to run something like this on corpus of documents I have and I'm not really a machine learning expert, any recommendation for a library/boilerplate/starting point?

I really enjoyed the article, it is (in my opinion) a nice case of curiosity-driven research.

I was wondering, what if someone would like to try a similar approach on images rather than text? What kind of image compression algorithms (if any) would be fit for use in this scenario?

Not a blog but an interesting source for many topics related to AI:

https://thegradient.pub/

While I agree that we need better metrics (and, more in general, a better academic system) I think that publishing fake research (committing fraud) is still way more severe than committing H-index hacking. The latter mostly translates into publishing lots of papers that present small advancements and does encourage scientists to work on topics where there is very little risk of failing (rather than chase challenging topics). These are have long-term negative implications for the systems and for our society as a whole, but, at the end of the day, it's still real research.

I think what you're describing is an utopia and likely will never come

Not to mention the 90% recall, meaning that the model is missing 10% of the bottles in the annotated data.

Cool blog post, cool tutorial, and great tool but its not going to win any contest where you need to guess an exact number.

I don't know a lot about chess but I have one question: isn't chess a solved game (in the sense that given a board state we always can compute the right move)? why use a neural network that can introduce, a very small percentage of the time, mistakes? I guess it is for performance reasons?

If you really want to get barebones you don't even need a static site generator, a template management library like handlebars.js or Jinja2 will suffice: load html template, fill the placeholders, save the results to file and you have your static pages.

Definitely the Sammezzano castle is on top of my list. Situated in Tuscany (near Florence) its owner had a crush for oriental culture. Thus, he decided to build this very peculiar castle in which each room follows a different arabic or middle-eastern architectural style. Pictures and more history here: https://www.homeinitaly.com/magazine/post.php?post_id=176.

Unfortunately it is privately owned and its kinda hard to register for a visit.

Maintaining 2 different apps (a native one and a web one) is costly in terms of time/resources. The point its not that developers "don't want" to make both, is that many don't have the resources to make both.

What you are asking is a well-known research problem (try opening Google Scholar and type in 'Android Sandbox') and I'm afraid no real solution exists at the moment. You can mitigate the problem in many ways and if you're concerned about your privacy I would suggest using mostly open-source apps (https://f-droid.org)

The article reads much like an 'Ode to Lynx' without providing real reasons on why one should use it.

Yes, you can do some sort of accessibility testing with it (and I'm sure it is an useful tool for a lot more corner cases) but I was expecting a lot more after reading the title

I guess stories will go down in history as the perfect example of a feature that was perfect in a single product (Instagram) and everyone tried to copy it in its own product, failing miserably.

Google Dataset Search is a good place to start:

https://datasetsearch.research.google.com

Unfortunately that's how most newspapers released the news, so I'm not surprised he's confused

I am aware of at least one competitor that also shows the price of the cheapest listing on ebay

Some very long calculation that a human would need time to parse, let alone compute, while instead a bot can do in milliseconds

Any suggestion for notebooks/examples that use interesting datasets? Yet another Twitter sentiment analysis its not exciting stuff for me.

As an academic myself I can somehow relate with Colin sentiment, not all work that is done in academia is impactful and/or novel, and indeed the current system rewards small incremental improvements rather than working on challenging problems.

However, I feel that there is academia and academia. I am lucky enough to: a) work with a supervisor that, during my PhD, encouraged me to ignore, and sometimes even challenge, the consensus view of the research community on the topic, believing that there's always more than one way to do things.; b) live in a country were universities are still mostly state-founded and so I do not need to constantly be hunting for grants.

Of course, there are compromises. I do have to spend some of my time on projects/papers that I hardly believe will ever have an impact outside of the academic world but that is pretty much the same as having to manage servers and customers, as long as it leaves me enough time to work on projects/papers that I enjoy.

Finally, I do believe that "small incremental improvements" is not the only way to do earn a living in academia. Being curiosity-driven, I consider myself more of a "generalist" rather than a "specialist" and I've been investigating rather diverse topics (e.g., https://news.ycombinator.com/item?id=22678471 and https://news.ycombinator.com/item?id=24361852). As I only have been doing this for a few years I have no idea whether my work will eventually be considered relevant enough to earn me a permanent position, but so far it has been quite well received by the academic community and I had no troubles in publishing it.

Author here. This is a first research we conducted to assess how frameworks for web-based desktop apps (e.g., Electron) are used in practice. It's a preliminary work that we plan on expanding in the future. Do the findings reflect your experience working with these frameworks? A re there other aspects we haven't mentioned in the paper that you deem relevant? What aspects of these frameworks you think could be improved?

I don't have a clear reference to point you to but in many CS research fields these number are equally bad. Often code is not even published and is a complete mess when it is.

For people who glaringly have too much money to even care about it is a fine really a punishment?

In fact we're going to see a great difference between those nations that invested the "money from the ground" to develop an economy and those who used it to buy soccer teams.

That's basically the whole point of "What the Most Successful People Do Before Breakfast"(https://lauravanderkam.com/books/successful-people-do/): do important things early in the morning, before they can be delayed by other commitments.

I recommend the book, its short and enjoyable.

How is this different from any basic web scraping guide?

I'd need to see the site before and after to confirm but maybe it is due to the faster page load speed? Tracking scripts these days are quite heavy and really affect loading times (especially on mobile)

It's yet another of the unfair advantages that China had over other countries, stemming from treaties signed decades ago that no longer make sense.

Glad to see it gone (and I'm not even an American).

Not exactly what you're asking for but a friend of mine is running a bunch of static analysis tools on these apps, looking for security/privacy warnings:

https://github.com/iivanoo/covid-apps-observer

I don't think these apps will be a success either but Immuni (the Italian contact tracing app) had more than 500k downloads in a single day on iOS and it's still in testing regime, available only in 4 regions out of 20