• BirdForum is the net's largest birding community dedicated to wild birds and birding, and is absolutely FREE!

    Register for an account to take part in lively discussions in the forum, post your pictures in the gallery and more.

A very comprehensive review of NL1242 VS EL1250 (1 Viewer)

PeterPS

MEMBER
No, deep learning (neural network) is not an "algorithm" based approach. You can't even tell exactly how it's reaching conclusions. (Never mind "AI", I agree that what people mean by that can be a mess)

Interesting review, particularly the suggestion to use the arguably redundant forehead rest as a substitute for extended eyecups. I'd like to try that myself. Thanks, range.

Also intriguing how machine translation gets tricky expressions like "fly in the ointment" just right, but stumbles over simple words like "mirror" (technical context?). I'd like to know the literal sense of this Chinese expression "wanting to eat up and down", which has many equivalents in other languages.

You cannot, but the person who designed the multi-layered neural network and trained it, and thus knows all its parameters, can. For you the NN is a black box, for the designer it's nothing but an algorithm that computes an output for each given input. The entire process is fully deterministic and under the control of the designer and there is nothing AI about it---the "machine" does nothing but what it was instructed/told to do.

Now back to the topic of this thread.....maybe range could clarify the sense of the expression "wanting to eat up and down", I was also wondering about it.
 
Last edited:

pm42

Well-known member
You cannot, but the person who designed the multi-layered neural network and trained it, and thus knows all its parameters, can. For you the NN is a black box, for the designer it's nothing but an algorithm that computes an output for each given input.
This is so wrong. Enter "neural network explainability" into any search engine and you will find the articles about it.

The entire process is fully deterministic and under the control of the designer and there is nothing AI about it---the "machine" does nothing but what it was instructed/told to do.
Wrong again. Because of dropout, because of the shuffling of the training set, etc.
Not to mention that once the network is trained, there is no easy way to predict its answer to new data.
 

PeterPS

MEMBER
pm42, post# 23


You have completely missed the point of my comments and rushed to post your insolent answer. Once a NN is fully trained the person who designed it and has all its parameters can compute its output to any given input (for "him", or anyone else who has all NN's details, the NN is a deterministic system/algorithm performing a known operation/function). Of course the output of the NN cannot be predicted with certainty in advance, but that's a different issue---my point is that it can be computed, once you have all the parameters of the NN, so there is nothing mysterious about the function performed by the NN, which does nothing that can be interpreted as AI.
 

PeterPS

MEMBER
Disagreeing with you is not being insolent especially because you do not know about the AI effect even when you are an obvious example of it:
https://en.wikipedia.org/wiki/AI_effect

You have your own definition of AI, algorithm and so on, think you are the only one to know about it and cannot stand a different opinion.
This is sad and I'm done talking to you.

This is not about having a different opinion but about the tone of your comment: "this is so wrong", "wrong again", etc. Even in your present post you tell me what I know and what I don't etc. Expressing a different opinion is of course always welcome, when it's done by a well-mannered person, but that's not your case.
 

tenex

reality-based
PeterPS, post #24 -- Thanks, now I do understand what you meant, though you could have explained your unexpected point better in the first place, as the term "algorithm" brought up the other point for me instead, as it did for everyone else here.

But yes, back now to Chinese idioms, or even the NL 12x42.
.
 
Last edited:

Troubador

Moderator
Staff member
Supporter
Gentlemen, that is quite enough about AI vs algorithms. If you wish to continue this debate, kindly take it to Ruffled Feathers. Thank you.

Lee
MODERATOR
 

John A Roberts

Well-known member
Australia
Another NL 12x42 vs EL 12x50 Review

Matt Cashell on the Rokslide site recently posted a comparative review with useful observations
see at: https://www.rokslide.com/swarovski-nl-pure-12x42-binocular-review/

n.b. A caution for some:
The start of the article includes a description of an unsuccessful stalk, the deer bounds safely away
If you wish, ignore the first 5 paragraphs, and go down to the comparative photo of the NL and the EL
The image shows each fitted with a 3rd party tripod mount


John
 

Attachments

  • NL x42 vs EL x50.jpg
    NL x42 vs EL x50.jpg
    102.2 KB · Views: 10
Last edited:

Users who are viewing this thread

Top