Theia FAQ’s with Jon

DrHow

Cloaked
Banned User
Premium Plus
Lifetime Premium
Corgi Lovers
Advanced User
Joined
May 18, 2018
Messages
4,341
Reaction score
7,603
I'm not really a countermeasures enthusiast (yet), but I came across this project the other day as a suggested youtube video. Really cool stuff you are doing here in SW Ohio, Jon and Co.!

A thought I had; you could use similar technology for lidar jamming. Specifically, while you've used a CNN to train for classification, it would be possible to use something like NEAT (NeuroEvolution of Augmenting Topologies) on top of a MLP to train a lidar jamming strategy. So you have a CNN classifier on the receive end, and for the response, all you'd have to do is pipe the error state of the gun back into the "response" network as an input for training. While a conventional NN takes advantage of gradient descent on a fixed topology, which limits solutions to local minimums, NEAT allows both the weights and topology to evolve. Theoretically, you could create a network which evolves the most optimized solution to defeat any given lidar. The only catch is that you need the error state of the gun, basically in real-time, which means you need any gun you'd want to train for. Since there is a response involved, and the lidar's reaction to that response is the desired outcome, you couldn't rely on recoded data. Although this has to be true for any lidar jammer manufacturer, they're just likely manually experimenting with responses/jamming patterns until they find something that works, with perhaps some degree of reverse-engineering the internals. As you've mentioned in videos, the main benefit would be the faster time to market with better, more effective jamming strategies than those found through manual experimentation. You could probably even do this with a vanilla network, but I have to assume that lidar manufacturers are using some pretty sophisticated anti-jamming, and if not today, will be, and a genetic algorithm like NEAT could likely brute force an optimized solution for any current or future anti-jamming strategy. It's all just a beam of light, after all.

Just thinking out loud here. Carry on. :)
^^^ This ^^^^

I had mentioned when Rai topic came out, that eventually Radenso could cut a deal to license some of their IP in the future for laser CM Partners (Or Rad create their own system). Assuming they stayed far ahead in this new order of thinking. Some of the above will be applied, just a matter of time. And will apply nicely to laser CM.

The above starts to explain what real AI is/does. Thank you for exposing some more of the reality. Welcome to RDF. As a former Oakwood (“under the dome” suburb of Dayton) citizen from the stork delivery to 22ish, happy to see Daytonion so well versed in the future. If you are a transplant, hit the Pine Club (if you have not already).
 

MachineLearner

Learning to Fly
Beginner User
Joined
Jan 3, 2020
Messages
70
Reaction score
203
Location
Dayton, Ohio
^^^ This ^^^^

I had mentioned when Rai topic came out, that eventually Radenso could cut a deal to license some of their IP in the future for laser CM Partners (Or Rad create their own system). Assuming they stayed far ahead in this new order of thinking. Some of the above will be applied, just a matter of time. And will apply nicely to laser CM.

The above starts to explain what real AI is/does. Thank you for exposing some more of the reality. Welcome to RDF. As a former Oakwood (“under the dome” suburb of Dayton) citizen from the stork delivery to 22ish, happy to see Daytonion so well versed in the future. If you are a transplant, hit the Pine Club (if you have not already).

Cheers!

Yeah, I suppose I'm technically a transplant to the Dayton area, but that was 25 years ago. I've come to see myself as a local at this point. :) Oddly, I've only been to The Pine Club once, and that was just a year or two ago. It was excellent, of course, but The Paragon, which shares a similar classic steakhouse aesthetic, is a little closer to home, and is my favorite haunt when I desire something in medium rare.
 

DrHow

Cloaked
Banned User
Premium Plus
Lifetime Premium
Corgi Lovers
Advanced User
Joined
May 18, 2018
Messages
4,341
Reaction score
7,603
Cheers!

Yeah, I suppose I'm technically a transplant to the Dayton area, but that was 25 years ago. I've come to see myself as a local at this point. :) Oddly, I've only been to The Pine Club once, and that was just a year or two ago. It was excellent, of course, but The Paragon, which shares a similar classic steakhouse aesthetic, is a little closer to home, and is my favorite haunt when I desire something in medium rare.
Paragon. Know that we’ll too. A young manager for the original owner of the Pine Club rustled up some investors and went south to Centerville. Don Berg. He was tragically killed in a car accident. I was very young back then, Don t remember for sure, but think his wife took it over. Not sure who has it by now, The Paragon has been a Centerville stable forever long time. Don copied the whole Pine Club operation, and expanded the size of the concept.

Pine Club has a steak named after my dad. At one point, same for Paragon. Filet, split, medium rare. With special recipe (dad’s) garlic butter in the side. Not on the Pine Club menu, but many thousands know about it. I seem to remember it did get on Paragon menu long time ago. The traditional sides from dad are screwed (stewed) tomato, salad with the famous pine club dressing (Paragon also copied that), and this uber thin battered onion rings (Pine Club original copied). With baked potato, with blue cheese dressing on top.

A third local south side steak joint that has stood the test of time, The Oakwood Club. I was doing very well several years ago.

Of course the best franchise, IMO, for coolest place great food and wine in Dayton... Cooper’s Hawk. We have them over here in Indy, and we are a member. Found out about them in Doral several years ago.

ok, back to AI.... I liked your brief above. Know there is much more to what you said. I think it is important for leaders in disruptive fields to teach others what is coming. True AI will be transformative in a good way for general population. Even kernel basic items towards AI, like Rai, will he major step forward. True AI is not stuck in some “we will be taken over by” .mil and giant evil tech companies AI efforts. Like IOT, it will touch our lives in many ways for the good.
 

MachineLearner

Learning to Fly
Beginner User
Joined
Jan 3, 2020
Messages
70
Reaction score
203
Location
Dayton, Ohio
Paragon. Know that we’ll too. A young manager for the original owner of the Pine Club rustled up some investors and went south to Centerville. Don Berg. He was tragically killed in a car accident. I was very young back then, Don t remember for sure, but think his wife took it over. Not sure who has it by now, The Paragon has been a Centerville stable forever long time. Don copied the whole Pine Club operation, and expanded the size of the concept.

Pine Club has a steak named after my dad. At one point, same for Paragon. Filet, split, medium rare. With special recipe (dad’s) garlic butter in the side. Not on the Pine Club menu, but many thousands know about it. I seem to remember it did get on Paragon menu long time ago. The traditional sides from dad are screwed (stewed) tomato, salad with the famous pine club dressing (Paragon also copied that), and this uber thin battered onion rings (Pine Club original copied). With baked potato, with blue cheese dressing on top.

A third local south side steak joint that has stood the test of time, The Oakwood Club. I was doing very well several years ago.

Of course the best franchise, IMO, for coolest place great food and wine in Dayton... Cooper’s Hawk. We have them over here in Indy, and we are a member. Found out about them in Doral several years ago.

ok, back to AI.... I liked your brief above. Know there is much more to what you said. I think it is important for leaders in disruptive fields to teach others what is coming. True AI will be transformative in a good way for general population. Even kernel basic items towards AI, like Rai, will he major step forward. True AI is not stuck in some “we will be taken over by” .mil and giant evil tech companies AI efforts. Like IOT, it will touch our lives in many ways for the good.

Pretty cool history there! I knew the two restaurants were related, I just didn't know how. I've been to The Oakwood Club countless times. For some reason, it was where my family landed when we were in the Kettering/Oakwood area, rather than The Pine Club. I think when my dad first came to Dayton for interviews, they took him to The Oakwood Club, so that became an early family staple.

As far as AI and Machine Learning, the stuff used in practice is really more simple than people perceive. Artificial General Intelligence is kind of what people fear when they hear the term "AI," but while that is being extensively researched, we're not very close to something truly general. In practice, Machine Learning for highly specialized tasks is where we are currently, be it image recognition, signal discrimination, natural language processing, creditworthiness (an area of focus for me), etc. Neural Networks in particular are actually little more than multiplication and addition, at least in the activation functions. The training is a bit more complicated, but still pretty easily grasped.
 

DrHow

Cloaked
Banned User
Premium Plus
Lifetime Premium
Corgi Lovers
Advanced User
Joined
May 18, 2018
Messages
4,341
Reaction score
7,603
Pretty cool history there! I knew the two restaurants were related, I just didn't know how. I've been to The Oakwood Club countless times. For some reason, it was where my family landed when we were in the Kettering/Oakwood area, rather than The Pine Club. I think when my dad first came to Dayton for interviews, they took him to The Oakwood Club, so that became an early family staple.

As far as AI and Machine Learning, the stuff used in practice is really more simple than people perceive. Artificial General Intelligence is kind of what people fear when they hear the term "AI," but while that is being extensively researched, we're not very close to something truly general. In practice, Machine Learning for highly specialized tasks is where we are currently, be it image recognition, signal discrimination, natural language processing, creditworthiness (an area of focus for me), etc. Neural Networks in particular are actually little more than multiplication and addition, at least in the activation functions. The training is a bit more complicated, but still pretty easily grasped.
Went to high school with one of the Oakwood Club owners offspring. They kept it in the family at least one or more generations. Not sure as of this moment who owns the place. It is under rated compared to the other icons. Many people have gravitated away from the Pine Club after ownership changes from Lloyd to Denny Hume (There were a few others in between that had a shot at buying it, but ventured out in their own.. Like buying Neil’s Heritage House or something else. Denny ended up with the PC. Some Daytonians have told me they do not like him, his personality, and how he runs things. They make good points. Many park themselves at Oakwood Club (or others). Denny has to be old enough to be looking for his PC successor, or someone already has taken the challenge.

Once someone sits and listens to those like yourself take them through AI, it normally is grasped. Some who do not like change will shake their heads. Thanks for sticking around here. Your distinctions will be added to the collective. :)
 

Eloi

Premium Plus
Lifetime Premium
Advanced User
Joined
Jan 13, 2013
Messages
4,663
Reaction score
4,538
Would Ai be sufficiently intelligent to correct his own error ?

Like " sorry this was not a true police radar , correction is in progress ..."
 

Windstrings

Premium Plus
Lifetime Premium
Advanced User
Joined
Jun 18, 2013
Messages
8,036
Reaction score
4,127
Location
Central Texas
I think You get what you get all in milliseconds time.
If you can call that "mulling it over!" [emoji28]

Sent from my SM-N960U using Tapatalk
 

MachineLearner

Learning to Fly
Beginner User
Joined
Jan 3, 2020
Messages
70
Reaction score
203
Location
Dayton, Ohio
Would Ai be sufficiently intelligent to correct his own error ?

Like " sorry this was not a true police radar , correction is in progress ..."

Yes, and no. In Theia's current implementation, probably not (guessing), but this is an excellent question.

Theoretically, Theia could implement crowd-sourced feedback into their network via wifi or bluetooth, retrain on this new data, and constantly be updating the classifier. The problem is that is introducing individual humans into the equation (who could intentionally or unintentionally enter incorrect data), and individual humans are quite fallible. Perhaps there could be an "opt-in" or "opt-out" to such a network, so for those that opt-out, they get purely Radenso-trained data, and for those that opt-in, they get a network trained on both Radenso data and crowd-sourced data. I suspect the crowd-sourced data would be better for identifying non-LEO falses; signals that RAI has not yet encountered, rather than new LEO radars, since there are a lot more false sources than legit sources, and legit sources is a slow-moving space (not many guns come out per year, for example).

My guess is that the strategy is to train on every known gun, and simply not alert if it is not one of those guns. That being said, for those of us that like to tinker more, it might be fun to play in an "alert on every unknown source" playground, and Radenso could build up a huge database of RF sources. Interesting thought, for sure.

All that being said, I just don't think it is needed. A classifier in this space really only needs to be able to identify radar guns, and it doesn't need to care about any other source. If trained correctly and thoroughly, it's quite unlikely to encounter a radar gun it cannot identify. The only unknown is really two emitters stepping on each other at roughly the same frequency. I don't know what RAI would do with such a scenario, but I'm sure they've thought of that.

I'll let Jon add his thoughts, though, I'm just speculating.
 

Eloi

Premium Plus
Lifetime Premium
Advanced User
Joined
Jan 13, 2013
Messages
4,663
Reaction score
4,538
Can Ai speculate ?
My guess is not yet ...i dont want to live in a machine-driven-world ...aka the max tragedy .
 

Jon at Radenso

Manufacturer
Premium Plus
Lifetime Premium
Corgi Lovers
Advanced User
Manufacturer
Joined
Oct 9, 2012
Messages
3,139
Reaction score
25,435
Age
32
Yes, and no. In Theia's current implementation, probably not (guessing), but this is an excellent question.

Theoretically, Theia could implement crowd-sourced feedback into their network via wifi or bluetooth, retrain on this new data, and constantly be updating the classifier. The problem is that is introducing individual humans into the equation (who could intentionally or unintentionally enter incorrect data), and individual humans are quite fallible. Perhaps there could be an "opt-in" or "opt-out" to such a network, so for those that opt-out, they get purely Radenso-trained data, and for those that opt-in, they get a network trained on both Radenso data and crowd-sourced data. I suspect the crowd-sourced data would be better for identifying non-LEO falses; signals that RAI has not yet encountered, rather than new LEO radars, since there are a lot more false sources than legit sources, and legit sources is a slow-moving space (not many guns come out per year, for example).

My guess is that the strategy is to train on every known gun, and simply not alert if it is not one of those guns. That being said, for those of us that like to tinker more, it might be fun to play in an "alert on every unknown source" playground, and Radenso could build up a huge database of RF sources. Interesting thought, for sure.

All that being said, I just don't think it is needed. A classifier in this space really only needs to be able to identify radar guns, and it doesn't need to care about any other source. If trained correctly and thoroughly, it's quite unlikely to encounter a radar gun it cannot identify. The only unknown is really two emitters stepping on each other at roughly the same frequency. I don't know what RAI would do with such a scenario, but I'm sure they've thought of that.

I'll let Jon add his thoughts, though, I'm just speculating.

This is a really great answer. You are bang-on right about why we are not enabling crowd sourced data from the get-go (would pollute our carefully supervised training set). But since we are open-sourcing the detector, I am totally open to the community loading their own trained network on instead of our "safe" OEM one.
 

MachineLearner

Learning to Fly
Beginner User
Joined
Jan 3, 2020
Messages
70
Reaction score
203
Location
Dayton, Ohio
This is a really great answer. You are bang-on right about why we are not enabling crowd sourced data from the get-go (would pollute our carefully supervised training set). But since we are open-sourcing the detector, I am totally open to the community loading their own trained network on instead of our "safe" OEM one.

Thanks, Jon.

By the way, I personally really appreciate the open-source attitude you're taking with this product. As someone who leverages and contributes to such projects both professionally and in hobbyist pursuits, it really is the best way to discover new solutions to old problems. Crowd-sourcing a "non-safe" network training set is a really interesting idea. You never know what's coming next, and who might discover an elegant solution to mitigate the problem.
 

Jon at Radenso

Manufacturer
Premium Plus
Lifetime Premium
Corgi Lovers
Advanced User
Manufacturer
Joined
Oct 9, 2012
Messages
3,139
Reaction score
25,435
Age
32
Thanks, Jon.

By the way, I personally really appreciate the open-source attitude you're taking with this product. As someone who leverages and contributes to such projects both professionally and in hobbyist pursuits, it really is the best way to discover new solutions to old problems. Crowd-sourcing a "non-safe" network training set is a really interesting idea. You never know what's coming next, and who might discover an elegant solution to mitigate the problem.

The way I think about RF has been altered forever by open source tools like gnuradio, so I feel it's only appropriate to open up the platform in like manner to others. And even from a selfish perspective in terms of making the best product possible - my lead DSP engineer said something I took to heart, "I'm not smarter than the entire internet."
 

PaulZy

Pax
Beginner User
Joined
Sep 20, 2018
Messages
76
Reaction score
70
Location
Phoenix Arizona
It will be interesting to see how manufacturers of radar guns respond to this new level of detection.

At first, the pool of AI users will be so small that not much will be done, as these new units will be expensive, atypical, and not cause much heartburn at the donut shop.

Eventually, the fuz will catch on and the wheels of the government will slowly reach out for a work around. But we, as evaders, are at an advantage, because we can respond faster, that is, unless laws are changed that would prevent the sales of these high level detectors. That is our biggest danger, as I can foresee it.

I certainly don't know much about this stuff, but it would seem that laser it their best defense to the general radar detector user market, as laser jamming defense is expensive and more tricky to employ correctly in a vehicle's front and rear.

Now there you go Jon, an integrated radar and Laser-jamming mounted unit - emitting a Jedi level force-field bubble of protection surrounding my vehicle. lol

I would pay a couple grand for that, easy!
 

DrHow

Cloaked
Banned User
Premium Plus
Lifetime Premium
Corgi Lovers
Advanced User
Joined
May 18, 2018
Messages
4,341
Reaction score
7,603
Yes, and no. In Theia's current implementation, probably not (guessing), but this is an excellent question.

Theoretically, Theia could implement crowd-sourced feedback into their network via wifi or bluetooth, retrain on this new data, and constantly be updating the classifier. The problem is that is introducing individual humans into the equation (who could intentionally or unintentionally enter incorrect data), and individual humans are quite fallible. Perhaps there could be an "opt-in" or "opt-out" to such a network, so for those that opt-out, they get purely Radenso-trained data, and for those that opt-in, they get a network trained on both Radenso data and crowd-sourced data. I suspect the crowd-sourced data would be better for identifying non-LEO falses; signals that RAI has not yet encountered, rather than new LEO radars, since there are a lot more false sources than legit sources, and legit sources is a slow-moving space (not many guns come out per year, for example).

My guess is that the strategy is to train on every known gun, and simply not alert if it is not one of those guns. That being said, for those of us that like to tinker more, it might be fun to play in an "alert on every unknown source" playground, and Radenso could build up a huge database of RF sources. Interesting thought, for sure.

All that being said, I just don't think it is needed. A classifier in this space really only needs to be able to identify radar guns, and it doesn't need to care about any other source. If trained correctly and thoroughly, it's quite unlikely to encounter a radar gun it cannot identify. The only unknown is really two emitters stepping on each other at roughly the same frequency. I don't know what RAI would do with such a scenario, but I'm sure they've thought of that.

I'll let Jon add his thoughts, though, I'm just speculating.
I suggested the above when Rai was introduced. And predicted this will happen at some point by Rad or others who go this route eventually. Thanks for confirming I was not in la la land.
Post automatically merged:

It will be interesting to see how manufacturers of radar guns respond to this new level of detection.

At first, the pool of AI users will be so small that not much will be done, as these new units will be expensive, atypical, and not cause much heartburn at the donut shop.

Eventually, the fuz will catch on and the wheels of the government will slowly reach out for a work around. But we, as evaders, are at an advantage, because we can respond faster, that is, unless laws are changed that would prevent the sales of these high level detectors. That is our biggest danger, as I can foresee it.

I certainly don't know much about this stuff, but it would seem that laser it their best defense to the general radar detector user market, as laser jamming defense is expensive and more tricky to employ correctly in a vehicle's front and rear.

Now there you go Jon, an integrated radar and Laser-jamming mounted unit - emitting a Jedi level force-field bubble of protection surrounding my vehicle. lol

I would pay a couple grand for that, easy!
Good future think. Let’s assume for the moment that radar makers are not all lazy butts. Or they get taken over by people like the person who owns Dragon Eye in Atlanta. They will respond.

In the laser CM world, the enemy comes up with new algorithms to thwart detection and jamming. So far, this punch counter punch world, ALP and TMG are winning (not without issues). In a future AI world where radar AI has to deal with real time Star Trek like “shield phasing“ codes, in this case vast shifting variable acting police radar, AI, combined with cloud sourced ”zero day” like algorithms examination (to counter what messes with AI trained sight picture); this will get into real time measures and countermeasures. Who knows if it will get this serious. Hacking today is already at this level. <milliseconds make a difference between compromise or not.

Why would all this advance to this level? Money. IF, IF, AI based CM (radar and laser) really make enough of a dent into the .gov money chest, these leaches will stop at nothing/spend whatever it takes (paid by taxpayers) to keep feeding their pig. They could pass banning laws. Right now, for most states, that can be risky politically. So, they will invest other people money in new tech measures.

I do not expect this dark look of CM world to completely come to pass. The money collectors will react. Could be a do nothing grunt, or all out warfare. Let’s hope LEO are so fat and happy with citation cash, that a few who get away are not worth LEO stepping up.
 
Last edited:

Windstrings

Premium Plus
Lifetime Premium
Advanced User
Joined
Jun 18, 2013
Messages
8,036
Reaction score
4,127
Location
Central Texas
So far, just keep high end detectors expensive keeps the RD user crowd small enough leos have zero problems finding fish to spear.

Imagine a world with a foolproof detector? Once the word got out... People don't even blink anymore to spend 1k on a phone.....
Who knows how many may cough up the money to guarantee no tickets.. Especially if laser detection were reliably countered too in the same box!

But I too expect that world will never happen before they will first start raising the technology of the guns.

Then just like laser, it will ge a tug of war as to who's got who's [emoji238].

Sent from my SM-N960U using Tapatalk
 

G37X Jockey

Premium Plus
Lifetime Premium
Advanced User
Joined
Jul 22, 2012
Messages
2,213
Reaction score
3,436
Location
Behind my Steering Wheel
I don't agree that radar gun manufacturers will conclude that the introduction of AI RD's will require a high-tech response.

Theia will be a much more sensitive DETECTOR, not a JAMMER. Our current RD's are already more than a match for C/O radar, and I/O will remain an effective countermeasure to all RD's - although less so for Theia if the range is anything like what's being suggested. I/O already exists and is very effective, so if I ran a radar gun company I would just emphasize the effectiveness of I/O rather than spend millions on new tech.

We're overthinking this and buying trouble.
 

MikePA

PSL +5
Intermediate User
Joined
Nov 5, 2016
Messages
1,251
Reaction score
1,916
Over and over and over and over again.
...and over and over ad nauseam.

  1. The number of drivers using a radar detector is a TINY percentage of total drivers.
  2. The number of drivers using a laser jammer is a TINY, TINY percentage of total drivers.
  3. The number of drivers using Theia will be a TINY, TINY, TINY percentage of total drivers.

Q: Why would any radar gun manufacturers invest ANY money to 'catch' people running Theia?
A: They would not.
 

Walter

Learning to Drive
General User
Joined
Dec 22, 2018
Messages
30
Reaction score
32
I was driving down the road and all of sudden I get full scale KA. I hit the brake hard and slow down. I wonder where is he. About 1/2 mile ahead there is a SUV running along and he pulls over. Yep, after I pass he pulled in behind me and hit the lights. He got my highest speed. You will not beat IO all the time, no matter what you run.
 

fitz4321

Running With Scissors
Premium Plus
Lifetime Premium
Advanced User
Joined
Dec 12, 2015
Messages
2,493
Reaction score
4,834
Location
Bay Area, CA
I suggested the above when Rai was introduced. And predicted this will happen at some point by Rad or others who go this route eventually. Thanks for confirming I was not in la la land.
Post automatically merged:


Good future think. Let’s assume for the moment that radar makers are not all lazy butts. Or they get taken over by people like the person who owns Dragon Eye in Atlanta. They will respond.

In the laser CM world, the enemy comes up with new algorithms to thwart detection and jamming. So far, this punch counter punch world, ALP and TMG are winning (not without issues). In a future AI world where radar AI has to deal with real time Star Trek like “shield phasing“ codes, in this case vast shifting variable acting police radar, AI, combined with cloud sourced ”zero day” like algorithms examination (to counter what messes with AI trained sight picture); this will get into real time measures and countermeasures. Who knows if it will get this serious. Hacking today is already at this level.
Why would all this advance to this level? Money. IF, IF, AI based CM (radar and laser) really make enough of a dent into the .gov money chest, these leaches will stop at nothing/spend whatever it takes (paid by taxpayers) to keep feeding their pig. They could pass banning laws. Right now, for most states, that can be risky politically. So, they will invest other people money in new tech measures.

I do not expect this dark look of CM world to completely come to pass. The money collectors will react. Could be a do nothing grunt, or all out warfare. Let’s hope LEO are so fat and happy with citation cash, that a few who get away are not worth LEO stepping up.

I think it would be naive to think there arn’t rardar gun manufacturers lurking here on RDF. Whether they care about Theia, who knows.
 
Last edited:

Discord Server

Latest threads

Latest posts

Forum statistics

Threads
95,433
Messages
1,452,287
Members
24,457
Latest member
Sgtmac
Top