A.I. Experiments: Visualizing High-Dimensional Space


[MUSIC PLAYING] DANIEL SMILKOV: Hi, I’m Daniel. MARTIN WATTENBERG:
Hi, I’m Martin. FERNANDA VIEGAS:
Hi, I’m Fernanda. Machine learning
is pretty complex. So we’ve been
experimenting with ways to visualize what’s happening. There’s a core concept
in machine learning called high-dimensional space. Here’s one way to wrap your
head around this concept. You can think about people
as being high-dimensional. For example, take
famous scientists. You can think about when they
were born, where they were born, their fields of study. Each of these is like a
dimension of that person. These dimensions become
difficult to untangle when you think about
different people, because someone might
be similar in some ways, but very different in others. MARTIN WATTENBERG: But
this is the kind of thing you can use machine
learning for. With machine
learning, the computer isn’t told the meaning
of these dimensions. It just sees them as numbers. And it sees each set of
numbers as a data point. But by looking across all
of these dimensions at once, it’s able to place related
points closer together in high-dimensional space. DANIEL SMILKOV: Here’s
a concrete example where words are treated as
high-dimensional data points. The important thing
to remember is that we haven’t told the
computer the meaning of words. Instead, we’ve shown it
millions of sentences as examples of how
words get used. Here is a visualization
of the results. We’re looking at
a subset of words that the computer
has learned about. Each dot represents one word. Each word is a data point
with 200 dimensions. Using a technique called
t-SNE, the computer clusters words together
that it considers related. And clusters
form-base the meaning, even though we’ve never taught
it the meaning of words. Here is a cluster of
numbers, months of the year, words related to space, people’s
names, cities, and so on. FERNANDA VIEGAS: We
can also look closely at smaller sets of words. If we search “piano,”
we can run t-SNE only on words related to “piano.” We get clusters of composers,
genres, musical instruments, and more. MARTIN WATTENBERG:
And this approach doesn’t just work from words. For example, you can
also treat an image as a high-dimensional
data point. Here’s a dataset
where lots of people wrote digits between 0 and 9. People write in
all kinds of ways. So the question is, instead
of us needing to manually code rules for all the
ways people write, could a machine figure it out
itself using machine learning? Each image is 784 pixels. The computer treats each
pixel as a dimension. Again, using t-SNE, it
clusters these images in a high-dimensional space. We’ve color-coded them so
that it’s easier for us to see what’s going on. And you can see groups of
digits clustering together. It’s learned something about
the meaning of these digits. FERNANDA VIEGAS: These
visualizations techniques we’ve been exploring can be
useful for all kinds of things. That’s why we’re working on
open sourcing all of this as part of TensorFlow
so that anyone can use these tools
to explore their data. [MUSIC PLAYING]

100 thoughts on “A.I. Experiments: Visualizing High-Dimensional Space”

  1. Remember, once Google refused a project releted to improve accuracy of drones and missiles using AI. Now amazon and microsoft took the project

  2. The sound and the video together… feels off. Where are the microphones? How did they record this? Is it just me?

  3. Open source are my two favourite words in this video
    And I really love AI, machine learning, & data visualization

  4. this is the tech version of how gypsies used to read people's palm and tell them about the future…..science is so far behind

  5. So that's why they collect user data on everyone. You have just been giving the answer to alot of what's going on.
    They are building a super AI.

  6. I can see this will eventually lead to an artificial consciousness as I'm able to discern those patterns as synapses and thoughts. Also star clusters…

  7. With these tools being open source we should all be concerned about the direction were headed in socially as a country.

  8. Guys you should contact DARPA, I can immediately see plenty of application for this in identifying groups (or individuals) of people who think wrong.

  9. As always, the problem is what to encode to a vector and what weighting to give each feature. That requires humans (thankfully).

  10. So the machine-learning algorithm is sorting words or letters by shape that appear in a certain area of the text close to each other. In a text of course related word are closer to each other, because they are used in the same context. What is so special about this kind of sorting algorithm? I don‘t get it.

  11. Oh, i thought you had begun to simulate high dimensional space in the physical sense of the word. How disappointing.

  12. Well we already access a higher dimension… Let's say you have a room jam packed with books… Well, digitalize them and save them onto a 32gb micro memory card. Boom. Physical space acquired thru the use of a higher dimension.

  13. This is the type of technology that will allow Google to manipulate the outcome of elections across the WORLD and fundamentally bring about any outcome they desire. Multi dimensional social engineering, this is wielding the power of gods

  14. You people are amazing , this will help me allot for my future project, keep up the good work I wish you the best endeavors.

  15. I’m watching this after Elon Musk talked abt computers pulling us into their dimension, oh gahd it’s happening.

  16. How do the AI know that months and names ect. are linked together? When they spoke about the AI grouping words together, I was thinking that words in an example sentence would be related to each other. For example, if I give the AI the sentence " I am hungry", it would cluster "I" next to "am" and "am" next to "hungry" because they saw those words being used together… It seems though I am mistaken. Can anyone explain how the AI finds out what are months and names and cities ect? Thanks

  17. I like that you're basically having a logical processor translate words to numbers, but are you giving it any principals of the language it's reading? Are the words and sentences the AI is interpreting grammatically correct?
    Have you done any tests with different ratios of writing to compare what it learns from say someone using English as a second language, to an Oxford English professor? Are the dialects similar?

  18. We develop the gis based esri arcgis application development team.we develop the customize map data and data analytic data

    Contact me on -sndp_ln

  19. Wait….What?, OMG was this a disclosure? Isn't this how the brain works, this is creepy as hell. So what's to stop them from creating a program {that is already is use} to go in and control these thoughts to suit their needs?

Leave a Reply

Your email address will not be published. Required fields are marked *