Using Google Cloud AutoML to Classify Poisonous Australian Spiders

https://shinesolutions.com/2018/03/14/using-google-cloud-automl-vision-to-classify-poisonous-australian-spiders/

by mattfrasernz • 8 years ago

107 41 8 years ago

Zhenya2• 8 years ago

Something like this could actually save lives. Take a photo of what bit you, then the app could provide an answer (with a confidence level) of "do I need to go to the hospital" and provide a few example apps for various matches with level of danger.

Thanks for sharing!

thrownaway954• 8 years ago

it's Australia... the answer is always "YES!!!"

zaarn• 8 years ago

The app could probably be optimized to the following:

    def spiderIsDeath():
      return True

Though surprisingly Australia does host some non-lethal animals and insects, including spiders. Most of the time people will be fine when bitten by spiders in Australia, treatment is quick in almost all cases.

jonathanpoulter• 8 years ago

I don't know if you get a chance to take a picture of what bites you most of the time... if ML could distinguish between different bites by just looking at skin, that would be incredibly useful.

Zhenya2• 8 years ago

Agreed with this also!

I wonder how different bites present themselves - if there is enough differentiation to provide a conclusive level feedback.

dbaupp2• 8 years ago

I don't think it is actually likely to save lives, because spiders very rarely kill. As far as I can tell, there's been one death from a spider bite in Australia in 40 years: https://www.telegraph.co.uk/news/2016/04/12/young-man-dies-a... .

thomasfoster962• 8 years ago

That’s probably because it’s fairly widely known (in Australia) that if you’re bitten by a red-back or funnel-web spider that you seek medical attention, as you would for a venomous snake. A quickly treated bite is almost never fatal.

The app helps people make the decision as to whether its a harmless spider or something that requires urgent medical attention.

bamboozled• 8 years ago

You'd be crazy not to get a second opinion at a hospital though right? "The app said I didn't need to see a doctor" sounds like dangerous territory.

Not saying a doctor is perfect, though I find they're often better than they get credit for.

tagh• 8 years ago

That is, if you haven't already squished it!

Speaking as an Australian, common spiders which aren't poisonous (e.g., daddy long-legs, huntsmen, little jumpy-guys) is general knowledge. As is the really bad ones (funnel webs, redbacks). If you're bitten by something else, you generally assume it's somewhat poisonous and pay close attention to how you're feeling.

This kind of approach would have more impact for snakes, whose venom can vary significantly -- although hanging around to snap a picture seems dangerous. Perhaps a suggestion system based on salient features?

lgats• 8 years ago

Now we just need to classify squished spiders.

michaelgreen4• 8 years ago

This is huge, and it's only an alpha. I begun reading about AutoML/Neural Architecture searches around ~year ago and something I've been thinking about is:

Why doesn't this just move the optimization problem? Aren't you now just optimizing your DeepRL network rather than the network you're trying to optimize?

CardenB• 8 years ago

Yeah but you basically leverage hierarchy principles to make the system better overall. At least I think that’s the idea.

nl6• 8 years ago

The idea of AutoML (in this case[1]) is to improve the NN architecture for a given type of problem.

In "normal" machine learning this is basically hyperparmater optimization for a given dataset (eg, the depth of a random forest, XGB parameters, the best random seed/jk )

In this case is tests different combinations of operators on a known dataset to see what performs the best. So it is optimizing the prediction network

(Also this isn't DeepRL, it's a deep neural network. I think that was a typo)

[1] https://research.googleblog.com/2017/05/using-machine-learni...

michaelgreen4• 8 years ago

Jeff Dean talks about AutoML using RL and in the paper " Neural Architecture Search with Reinforcement Learning" it also talks about this.

Also it seems different from more traditional hyperparameter optimization because it makes novel cells. So the structure of the network isn't limited to our existing library of layers/cells.

https://arxiv.org/abs/1611.01578 https://youtu.be/HcStlHGpjN8?t=2073

dzhiurgis• 8 years ago

If pictures are already from Google, wouldn't it be simpler to post image to Google's Vision API and then grep results against the list of poisonous spiders?

juskrey• 8 years ago

Isn't that something available in, say, Mathematica in 3 lines of code? https://wolfram.com/language/11/image-and-signal-processing/...

nl6• 8 years ago

No.

The example there is 7 lines (counting the NN description as one line). That's using a (easy) pre-existing dataset too, and a primative neural network.

That's roughly the same as in Python using something like the fast.ai library. I think that comes for 4 lines (not including data wrangling or inputs):

data = ImageClassifierData.from_paths(PATH, tfms=tfms_from_model(arch, sz)) learn = ConvLearner.pretrained(arch, data, precompute=True) learn.fit(0.01, 2) log_preds = learn.predict()

See [1]

Also note that this AutoML version uses zero lines of code.

[1] https://github.com/fastai/fastai/blob/master/courses/dl1/les...

neurostimulant• 8 years ago

Is it possible to export the trained model from AutoML for offline use?

milesokeefe• 8 years ago

They haven't come to a decision as to whether that will be offered AFAIK. I really hope they decide to allow downloading the model.

rrmoelker• 8 years ago

Does anyone know a bit more about the pricing? All I can find online is Google learning infrastructure cost.

skeleton• 8 years ago

I'm surprised someone is yet to point out the error of using the word poisonous instead of venomous. I suppose it's not important to the article.

socceroos2• 8 years ago

Although, if you do decide to nibble one you may find it is indeed disagreeable.

gajju35882• 8 years ago

Auto labeling would be way forward for Supervised Algorithms. Get some data to annotate from your team, and tag rest of them using auto-labeling. https://dataturks.com/ could be such player, Not sure how will these survive in front of Google.

nl6• 8 years ago

There's no such thing as auto-labeling.

Data Turks is manual labeling.

There is active learning[1] and related algorithms where you trace the boundary of your classifier and pass examples along that boundary to be manually labeled (as they are the ones the classifier is most unsure about).

But there is nothing "auto" about this - it's just being smart about where to deploy the manual labor.

[1] https://en.wikipedia.org/wiki/Active_learning_(machine_learn...

gajju35882• 8 years ago

Lets say we want to create a labeled data for text summarization for medium articles. Could the highlighted part be used as summary, its not auto labeled per se, but can be a proxy and passed to labelers to verify/edit.

thejosh• 8 years ago

How long does it take to get access to AutoML?

bastih• 8 years ago

I like the content of this blog post, but the use of stock photos was really off-putting and I had to almost force myself to keep on reading.

Using Google Cloud AutoML to Classify Poisonous Australian Spiders

Related Stories