Name: TFラティス。単調性でMLを制御する (TF Dev Summit '19) (TF Lattice: Control Your ML with Monotonicity (TF Dev Summit '19))
Uploaded: 2021-01-14T10:28:36.000Z
Duration: 5 min 7 s
Description: VoiceTubeの動画で発音を聞きながら英語表現を覚えよう！学べる英語：

And I'm excited to share with you about TF Lattice.

of our production models, where we care about the model

that we can guarantee it's working sensibly everywhere.

how you can control your models to have monotonicity.

And we have plenty of data here, 24,000 training examples.

And there's more features in the dataset.

We're just going to look at two, so that we

And what I'm showing here is the actual training data.

So the two features are, are you single or are you married?

And how many months has it been since you paid your bills?

So what you can see is that while there's 24,000 samples,

there's not that many people who haven't paid their bills

So we're starting to get a bit sparse in that part

And what you're seeing here is the training data,

in the mean and the confidence interval of the actual training

So what happens when we go and train a model?

Sort of visualize what's going to happen.

It doesn't matter if you're single or married,

it really matters how long it's been since you paid your bills.

paid your bills for seven months than six months

that you're going to get a higher credit score.

And this is one of the reasons that some people don't like AI.

If we had the full 24 dimensional features

to be pockets of the space where this sort of thing

realize that we're getting this sort of bad, strange, possibly

But any of your standard regularization techniques

It's going to be hard to really fix this without hurting

lets you have monotonicity as a regularizer.

You can say when you put together your features,

for this feature it should only hurt my model,

it should only hurt my score if I haven't paid

And so you can see here, this fixes that trend

We actually get slightly better test accuracy,

but we've solved the exact problem that we need to solve

and we now have a guarantee on what the model is doing.

And if we had 100 features, it would similarly

of 100 dimensional space, this was working correctly.

Well, under the hood, the kind of function class it's using

are lattices, and these are just interpolated lookup tables.

You've seen these in the back of your math textbooks.

You can find them in tables from actuary in 1960s, et cetera.

So in a one-dimensional space, these are simply

can represent high dimensional functions also

Here's an example with just two features.

We're building a function with 10 lookup table parameter

are trained using empirical risk minimization.

It's all the same training that you would see with DNN.

It's just the parameters now represent what's

字幕リスト動画再生

TFラティス。単調性でMLを制御する (TF Dev Summit '19) (TF Lattice: Control Your ML with Monotonicity (TF Dev Summit '19))

sort

constant

guarantee

realize