What is crystal ball?

I have seen some comments by some of the answered referring to crystal ball, been bugging for days. Does it actually suggest you a solution? Or is it just my hallucination? Can it be only obtained by certain people? I feel it as some kind of superpower ?♂

[Torsten]

You mean John D'Errico's foggy crystal ball ?

Here it is:

https://www.google.de/search?q=crystal+ball+%26+magician&tbm=isch&source=iu&ictx=1&fir=5O7I5w28kx6zqM%253A%252Cu0nKGxHs6fSISM%252C_&vet=1&usg=AI4_-kTNUAVWTW-KEfmPaSGH4kHK1wyl2A&sa=X&ved=2ahUKEwjr15qHjqrjAhVFDuwKHet4D10Q9QEwBXoECAIQCg#imgrc=5O7I5w28kx6zqM:

[madhan ravi]

Not sure Torsten , kind of the one seeking for.

[Torsten]

Explaining jokes destroys them :-)

[madhan ravi]

;)

[Stephen23]

"the link you gave just opens a picture of crystal ball."

Yes, because that is the correct meaning of "crystal ball". See also:

https://en.wikipedia.org/wiki/Crystal_ball

"Do you mean in File exchange?"

Did someone upload a crystal ball simulator onto FEX? That would save a bit of a space on my desk! This big orb just collects dust, burns my papers, and never seems to be very reliable...

"Does it actually suggest you a solution?"

If you owned a functioning one (i.e. one that supports clairvoyance, scrying, etc.) then it would let you understand OP's who do not provide enough information for us to help them effectively.

It might help with solutions too, but the model I have does not have the "solutions" feature, which unfortunately is only available via an expensive upgrade: first paying the medium to transfer the upgrade from the other realm, then flashing the ectoplasm, restarting at midnight at some crossroads, and finally pressing the "power" and "mode" buttons when the prompted by the startup visions...

[madhan ravi]

Haha yes for sure it might be expensive.

[John D'Errico]

I admit, my crystal ball is always foggy. I've tried ordering some crystal ball cleaner, to no good result. Lately, I've tried using Windex. It works no better, but at least it is much cheaper. So then I realized the problem is not the crystal ball, but my dirty eyeglasses. Once I cleaned them, I can now answer questions before people even ask them.

In fact, I may have been the person who originated the comment, to indicate a question where the information provided is wildly insufficient to answer.

[madhan ravi]

:-)

[Tim Fulcher]

Hi Madhan,

My reply is a bit late and I hope you've sorted your problem. But if you haven't the crystal ball function is a probability distribution function used in high energy physics. It's a combination of a power law function and a Gaussian. Wiki has a decent description. It looks, typically, like this::

[Tim Fulcher]

I have synthesised a C-B function but not yet fitted:

Answer 1

[Image Analyst]

It's a Toolbox, but as of the moment, the only person outside the Mathworks staff who has it is Walter Roberson. He seems to be the only beta tester in the general public that the Mathworks trusts to evaluate it.

The times it will be evident that he has it, and has used it, is when a poster posts a question that is horribly missing of information required to give a meaningful answer (or gibberish), yet Walter posts a brilliant answer and the poster replies "Yes, that's it!!!" The rest of us are left scratching our heads thinking "How could he possibly know all that? Those requirements were never stated." The Answer is : the Crystal Ball Toolbox. It allows the user to see the poster's screen so that they don't even need to pose a complete answer or post any script or data.

A related toolbox, still in alpha, is the Mind Reading Toolbox. I suspect this is even more powerful as it lets you get into the mind of the poster even before the poster has written any code or even formulated a comprehensible question. I again suspect that Walter is the only one allowed to test this one.

[madhan ravi]

Interesting you even addressed the fact about Mind Reading Toolbox, which I was thinking to ask about!! , will you be able to access it anytime soon ;) ?

[Steven Lord]

No Mind Reading Toolbox, but some of our users are doing interesting work with MathWorks products and people's brains. Two such stories:

Helping a paralyzed man move his hand.

Letting a robot know if you think it made a mistake.

[Jan]

20 years ago I've seen several announcements of the dwim operator as part of the coming Matlab version in CSSM, or at least an undocumented feature('dwim'). "Do what I mean" would be such important tool. Meanwhile I'm afraid, this was vaporware only.

[madhan ravi]

Steven those are amazing works right there , never underestimated the power of MATLAB. Jan I would have been 1.5 years old before 20 years, could you share any links that talks about that feature?, would be great.

[Jan]

It is hard to find links to the old CSSM newsgroup. What a pitty, because it contained a lot of gems. Try e.g. https://groups.google.com/forum/#!searchin/comp.soft-sys.matlab/dwim|sort:date/comp.soft-sys.matlab/vHY0agQ9-E0/p6UXhcicf8sJ , or start at https://groups.google.com/forum/#!forum/comp.soft-sys.matlab

The DWIM feature has been suggested repeatedly for question with the pattern:

"I have some code, which computes <abc> , but I want to get <xyz> instead. What do I have to change? A friend told me, I should change everything, but my professor does not agree."

[Adam Danz]

@Steven, it's not as amazing as the links you shared but I use Matlab every day in a project that will use behavior and neural responses in the visual system to try to estimate the planned trajectory of macaque monkeys as they steer along a memorized and invisible road in a virtual environment. Does that qualify as crystal ball technology?

[Image Analyst]

It's close, but in yours the user (monkey) has already seen the situation in advance and is trying to reproduce it, while in the Crystal Ball Toolbox, no viewing of the remote computer screen in advance is needed.

Nonetheless, your code could have reapplications if you can bundle it up and post it to the File Exchange. For example, show your monkeys pages from Shakespeare and then afterwards have them try to type out a brand new Shakespeare play and have them test against the deep learning networks that are trying to do the same and see which did it better.

[madhan ravi]

Thank you Jan :)

[Star Strider]

Following up on Adam Danz’s Comment, in the early 1990s, I was working tangentially with some physiological psychologists who needed my expertise as a physician (endocrinologist) with their work on human factors workolad studies. I got interested in this, and asked if it was possible to determine the task a person was performing on the basis of EEG activation patterns. Apparently no one had, and when they asked if I wanted to explore that, they said they had some archived EEG records from other experiments that would work to test my hypothesis. We did short-time Fourier transforms over time windows of all 20 leads over the full study, then took alternating records from each lead as training and test sets. Using a relatively primitive simple linear discriminant analysis and PCA (BMDP software), we were able to classify the test data (determine the task the person was doing) with about 66% accuracy, with a very low p-value. (I imagine that would be much easier and more efficient with MATLAB now.) We published that in 1995, at which time only one other group in the world (Pfurtscheller) was doing this sort of research. I have since read of significant advances in this area, although I have not kept up on it. Not quite ‘mindreading’ but close!

[Rik]

That sounds really cool. I've heard of some study with MRI where they showed participants images and recorded the response of the visual cortex. Then they asked the participants to imagine a scene and recorded the response again. The images were very fuzzy and you need an MRI (or more likely an fMRI), but they got really nice results.

[Image Analyst]

Rik, yes that was presented around 5 years ago by a neuroscientist at Univ. California, Berkeley. I saw his keynote talk at Electronic Imaging (Jan. 2014 I believe). They basically collected a bunch of brain patterns of subjects who were shown images while in an fMRI. Scenes of beaches, babies, cars, of whatever. Then they trained a deep learning network with those brain patterns, and when they showed subjects random/unknown (to the researchers) images, the trained network was able to predict what the subjects were viewing with an impressive accuracy.

If they can do this level of mind reading today, imagine what they'll be able to do in 100 or 500 years. Kind of fascinating yet scary at the same time.