waleedka

Good point. If you haven’t read “They’re Made out of Meat”, you’ll find this short story very amusing:

http://www.mit.edu/people/dpolicar/writing/prose/text/thinki...

Same here! I was going with the assumption that standard environments are going away since they haven't been updated for years.

I think you got it backwards. Which seems to be a fairly common mistake because, intuitively, we tend to associate the amount of work to the amount of value.

In Bitcoin, and other PoW blockchains, the protocol dynamically adjusts the difficulty of mining a new block such that it requires more work if there are more miners. Which also means that it requires less work if there are fewer miners, which is why it was easy to mine Bitcoin on a laptop in the early days but now it requires specialized hardware.

The causality path works like this: Price of the coin goes up -> it becomes more profitable to mine -> more miners jump in -> difficulty goes up -> cost of mining increases. This continues until there is an equilibrium.

One idea you might want to consider is to create a global function, let’s call it h() and pass all your human readable strings through it, like so h(“hello world”). This function, simply returns the same string. This is more explicit than relying on quote types. It also allows you to do interesting things such as logging everything to a text file and running a spell checker on it, or checking for wrongly encoded string, ...etc.

Actually I prefer standardizing on double quotes and I do that in my code on a regular basis. I know some people use a mix, sometimes single quotes and sometimes double quotes, which is fine, but I prefer consistency in this case.

You can now do it quicker with CTRL+Shift+V (Windows) and Option+Shift+Command+V (Mac). Both will paste without the formatting.

Does Swift have a large community beyond iOS apps? Last I used it was years ago, right after it was introduced by Apple. I'm curious if it has found growth in other areas.

> Construction of this data set consumed millions of compute hours

It's great to see such datasets shared publicly. I hope it achieves the intended purpose and spurs innovation in protein folding.

I’m very curious about the use of such cloud finder. But, to answer your question, the Mask RCNN model would be useful if you want to identify individual clouds. As in, find clouds surrounded by empty sky. Or, for example, you want to find clouds that look like a puppy or or a sheep. On the other hand, if you don’t care for identifying individual clouds and only want to find the pixels that belong to the ‘cloud’ class, then a semantic segmentation model would likely do better.

Also, while I don’t know much about your use case, using DenseNet seems like it might be an overkill since you only have two classes, cloud and sky. A lighter network might give you better results, especially if you don’t have a lot of training data.

I don't see this flag on my version of Chrome (64.0 on Mac). Are you on the beta branch?

Yes, totally agree. TF is solid and it’s still my goto tool, but the API has a very low elegance score, to put it nicely. Clearly the team has a lot of brillian engineers, but they’re missing the high level sense of design (yes, APIs do need good design as well) that puts everything together beautifully.

Not only the language, but have at least one line that says what this is. I can guess that it’s a hosted notebook service, but I can’t tell how it’s different from Jupyter Notebook. Is the difference in the language it uses, or are there other differences?

It gets worse, actually. People often, accidentally, click on the pen icon after highlight some text and end up highlighting that text permanently for all readers. And you, as the writer of the post, have no ability to remove that highlighting.

I do it to mark where I'm at in the post if I need to switch to something else temporarily. Makes it easier to continue from where I left of.

Same here. I had a movie available to me on Prime but I couldn't broadcast it to Chromecast and I didn't want to watch it on the small phone screen, so I had to buy it on YouTube.

Thanks! Done. Although I hope I was clear that my comment is general and not specific to the topic at hand. I just think people should share the facts but be careful about making assumptions.

This is a general comment not specific to the topic at hand. If you have facts to share then do so. But if you 'feel' that someone has an ulterior motive and you share it in a public forum then many people might take it as fact or an anecdote. Some people might even repeat it to their own friends. By doing so you risk harming the reputation of someone who may be innocent of that accusation.

Let me demonstrate with an example: I 'feel' that you wanted to join GOAP because you wanted to hang out with attractive women. And I 'feel' that 500 Startups didn't respond to you because they recognized your sleazy intentions. And I feel that you got upset and this is your way of getting revenge. Now, obviously, I don't really feel that way. But I just wanted to show how easy it is to throw dirt at someone.

Let's stick to the facts, please.

disclaimer: I was in 500 Startups first batch.

This architecture is optimized for accuracy rather than speed. The official paper reports 200ms inferencing time per image on a GPU. This implementation is likely a bit slower because we use Python in a couple of layers. This is easy to optimize, but we haven't gotten around to it yet.

With that said, there are a lot of things you could do to make this much faster. For example, use ResNet50 instead of ResNet101. You can also reduce the number of anchors or the number of proposals to classify, and that should improve performance significantly at the expense of a little loss in accuracy.

Evaluation code against MS COCO is included in the repository, both for bounding boxes and segmentation masks so it should be easy to run (but takes a long time).

We should publish more details, though. Thanks for bringing it up. Our implementation deviates a bit from the paper (as mentioned in the documentation), and optimizing for COCO was a 'nice to have' rather than being the main objective. We got pretty close to the reported numbers (within 3 to 4 percentage points) but that was with half the training steps compared to the paper. We'll try to add more details over the next few days.

They're in the "Releases" section. It's a 250MB .h5 Keras file.

[spoiler]

I interpreted that as fighting against other civilizations that are trying to stop your progress and protect the universe.

Ha! Newbie. You still have 99% to go :)

LOL. I'm guessing you're sarcastically jabbing at Slack for the known issue that their desktop app uses a lot of memory.

Tip for the project lead: A quick way to lose half of the early adopters is to release a library without Python 3 support. Luckily, fixing this is easy and shouldn't take more than a day or two. And the ROI is high if your goal is wider adoption.

The nice thing about Python is that it's intuitive and feels natural. To that end, this syntax for defining functions is nice, and I would love to see it implemented in native Python:

    f(x) = x^2 + 1

But most of the other ideas are less intuitive than their native counterparts. For example, I think that the native form:

    3 in [1,2,3,4,5]

is clearer than:

    [1,2,3,4,5] /contains/ 3

I'm not sure I'm following. Are you saying that using supervised learning is the wrong approach here? What would you use instead?

Author here. This first part is simple by design. It's targeted to those getting started in the field.

Author here. No, it's not related to Udacity. It's my own side project.

I'm intrigued! How do you interact with these images right now? Do you display it on screen and rotate it with the mouse? Can you elaborate a bit more about how would a VR/AR solution add value here other than being a fancier solution?

My understanding of the upper age limit is that it's the "natural" limit (i.e. if you don't take medicine and gene therapies and such). The goal of these new anti-aging therapies is to raise that natural limit.