Anyone Using Artificial Intelligence With Their Gold Detectors?

[Ben201000]

Welcome,

I was wondering if anyone if using any form of artificial intelligence in their gold detectors? 

I'm an artificial intelligence (AI) programmer. Last night I was watching Aduk gold on Netflix and suddenly wondered if anyone was using AI? A quick google didn't return much. 

I'm not sure how familiar everyone is with AI, but essentially since around 2015 it's blowing away peoples expectations year upon year. Simply put, traditional computer programs require humans to code in rules that then lead to a result. AI does that backward, it takes the results and creates its own rules to get to that. 

To do this with gold would require the creation of training data to feed into the AI. I don't know anything about gold detecting but I imagine you'd bury some pieces of gold and go over it with the detector, then save the waveforms (or equivalent) onto a computer. These become the gold samples. Then also bury things that most often give the most false positives compared to gold, and save those waveforms. 

The exciting thing compared to a few years ago is that not a huge amount of training data is needed. It's possible to take huge AIs trained by Google and then teach them the new gold samples far quicker with a lot higher accuracy. The process of training is basically the AI guessing 'gold vs not gold', and over millions of iterations it starts to learn. It then can be saved and used in the field to give a percentage estimation of how likely it thinks something is gold. 

I don't know anything about gold detecting but this is how I would see it used practically. Am I right in thinking that a fair amount of time is spent digging up false leads? If it's not and most of the time is spent surveying the area then the AI isn't very useful. But if there is a lot of time digging up false leads, then if the AI could save someone digging up 90% of the false leads would this create a lot of value? 

I'm really interested in any thoughts that any of you have regarding this. 

Cheers

Ben

[klunker]

  I have often thought that some intelligent person (that leaves me out) could build a pulse inductive detector that could be "trained". Nearly all detectorsist will discriminate targets to some extent by the type of target response heard through the headphones. It's not a very smart thing to do as I can guarantee you will miss gold but you just have to decide if your time is more productive covering acreage or digging a target that sounds just like the last 500 square nails that you dug. I have fooled myself may times when I knew it was trash but it turned out to be a nugget and where is was thinking of ways to spend the proceeds from my sure thing undug nugget but only to be rewarded with another square nail. However there are times when I know for certain that a target is a gold nugget. Can't say if it's a very "trained" ear or if I'm weird,

[Redz]

I see this as being particularly useful for problematic hot rocks, especially if you can grab your sample on the spot. Perhaps even as a smart ground balance.

[GB_Amateur]

3 hours ago, Ben201000 said:

I was wondering if anyone if using any form of artificial intelligence in their gold detectors?

Unfortunately some don't even use natural intelligence.  😎

3 hours ago, Ben201000 said:

The exciting thing compared to a few years ago is that not a huge amount of training data is needed. It's possible to take huge AIs trained by Google and then teach them the new gold samples far quicker with a lot higher accuracy. The process of training is basically the AI guessing 'gold vs not gold', and over millions of iterations it starts to learn.

This paragraph seems to contradict itself.  "...Not a huge amount of training data is needed."  Then the next sentence:  "...over millions of iterations it starts to learn."  (emphasis mine)

I undetstand in machine learning that if a lot of data has already been 'mined' (no pun intended, but it's a good one here 😁) and properly formatted, letting a computer learn from those data can be relatively quick.  But what you are describing doesn't sound like data that can be quickly nor easily collected.

One thing that even regulars here sometimes overlook is that the customer base for metal detectors can't be compared with many consumer products (such as cellphones).  What might be obvious R&D funding expenditures for products in high demand can be prohibitive in limited product markets.

 

[Clay Diggins]

Welcome to the forum Ben. I'm a fellow programmer and can probably show the basics of where the road blocks exist in the current technology. I am not a detector expert but there are a few on this forum, hopefully they will chime in. Please don't take offense at my comments. They are directed to all readers and aren't meant to denigrate the programming work you are doing today or the current state of "AI'.

First off a pet peeve- AI since 2015? Nothing like changing the definition to achieve the goal. You certainly mean neural network Weak AI? Alan Turing is rolling over in his grave.

Real Artificial Intelligence doesn't exist - it's only hypothetical at this point. Real Artificial Intelligence wouldn't require humans feeding training data to a program. What the public is led to believe is that intelligent Strong AI exists then they are offered Weak AI as proof of concept. We were employing Weak AI back in the late 1970's at TI - nothing new but the name.

To answer your question the "AI" you are referring to is used in several metal detectors today. You can start with automatic ground balance. Signal acquisition and processing is where most of the development is centered today. GiGo applies to metal detectors just like every other real world system and there is a huge amount of garbage in these signals.

As already mentioned humans perform this sorting function with the currently available detecting technology based on individual experience. Also noted is that even the best detectorists either dig every target or admit to missing gold if they don't dig every target. In other words human sorting of the data received by the detector is still very poorly developed even among the best operators. If there were true artificial intelligence the best we could expect is the same level of data sorting. I have doubts about the usefulness of an "AI" system that has no more capability than the existing human operators. Metal detecting isn't a button pushing job at a factory.

With the current BFO, VLF and Pulse detecting systems it is not possible to directly correlate signals received with the substance being detected. Not enough of the right kind of data. The use of slot filters can eliminate unwanted signals but those will also miss many of the target metal signals as well. As you can see from previous comments some believe a processor enabled slot filter amounts to AI. Thus my previous comments.

A different physical data collection device is needed to distinguish individual metallic signals.  At present no such system is available to the detecting public but you can bet some smart folks are working very hard on that goal. Of course once a system is designed that can discriminate among metal types there will be no need for AI to parse those signals for the operator.

I understand where you are coming from. My first thought when I picked up  a BFO  back in 1973 was that these machines need more intelligent processing to make sense of the signals created. I too misunderstood what was meant by a "metal detector".  A lot of work has been done already towards the goal of cleaning up the signals from the existing systems. The best we can do now as programmers now is sit back and let the engineers discover a working physical system to get the data that will allow true metallic discrimination.

 

[Norvic]

Recently downunders ABC had this bit of news.

Google fires software engineer who says AI chatbot LaMDA has feelings - ABC News

Suspect Ben201000 your on the money, the future lies in that direction no doubt why not our detectors, if AI can handle the noise (ground, EMI etc etc) we`ll have more depth/sensitivity. 

[klunker]

  Mr. Clay Diggins;

 

I couldn't have written that better myself.  Probably couldn't have written that at all.

[Mike_Hillis]

Hi Ben,

You should buy a metal detector and join us.   

All I want to see/experience/hear is the object's density as compared to a standard, with an audio output that tells me how close the signal response is to the standard.

HH
Mike

 

 

 

[1515Art]

I and scores of detector users await your success, as pointed out the challenges are many the variables countless, just the gold alone returns different signatures depending on size, orientation, composition, density, surrounding minerals and trash, oxidation… the list goes on and on. But if you could pull it off the line to you’re door would be very long, and why large companies like Minelab invest millions of dollars and thousands of hours on research seeking the answers to those questions as we await the next new thing. Best of luck, for sure.

[geof_junk]

The gold varies in size (no two are the same shape or size), content of other metals in it, the change of ground, moister in ground, and most of all how the operator uses the detector.  Such as speed, attention, consistency of coil height and over lapping. A big job for humans (risk assessment of leaving gold behind) bigger job for AI.