Hacker News Reader

Comment by mhh__ | Hacker News Reader

Actually - I sort of want to work in defense for a few years to find out, that and the sheer amount of tech in a modern missile (the fact that they would likely be used is why I didn't apply) - I reckon most the algorithms here are basically finely tuned signal processing and other non-"AI" processes.

There's a tendency to think that everything that isn't completely ab initio like machine learning is bad or inelegant - I often fall for this trap myself, but to just get the job done you can get a very long way with "dumb" algorithms and a practically infinite budget.

View on HN

Algorithms that don't use machine learning aren't dumb. The point of machine learning is you need to implement some function, can't derive it directly from first principles, but have lots of examples of correct inputs and outputs and can use that build a function estimator instead. But in a whole lot of quite sophisticated and intelligent applications, you can derive the functions from first principles. Not easily. One of the projects I spent the first few years of my career working on was the common image formation processing for spy satellite collections, and you need to take into account special relativity, orbital mechanics, curvature and velocity both of earth and of the arrays, midnight crossover in time keeping, temperature calibration and the impact on reported voltage detection of each sensor cell, parallax effects of rapid altitude changes, polarization of light. There's a ton that goes into it, but we know the physics and don't have to use statistics to make educated guesses. We can compute "given voltage levels x1, x2, ..., xn on sensor cells y1, y2, ..., yn at times t1, t2, ..., tn, that is what we were looking at" exactly.

Mind you, I'm talking basic level 1 transformation of raw data streams to human-intelligible images. Once you get into automated object recognition, that's when we start to use machine learning, but the algorithms upstream of that are still plenty smart.

GPS is a good example. Heavy math. Lots of physics. Not dumb at all, and you'd never get sub-1m accuracy with ML-GPS.

Once you get it right (& get a lock) it works every time, instead of occasionally deciding you are in Nome.

As a theoretical physics student who can actually write good code, this is the kind of thing I want to work on - unfortunately where I live you basically have a choice of making things and making money

Actually, when I was on this team, I think I was the only person with a CS background. Almost everyone else was physicists and electrical engineers.

> who can actually write good code

If you're going to make this statement, you need some skin in the game by putting your GH in your bio.

I don't have the balls to boast about my code so there is no GitHub link in mine :). There wouldn't be much to see anyway.

I work for the D language foundation, and they seem to trust me (god help them!)

It pretty much is the case.

Part of it is that predictability is a desirable feature in these systems, but also that problems like the one described in the article you don't really need things like ML. The majority of control problems like this are surprisingly straightforward. They might be complex, in the sense of having a lot of variables, but the physics involved is well understood and can be modelled using traditional techniques.

The progress of self-driving cars is a good example of this. I can remember seeing expeimental self driving cars many years ago, but always going round mostly empty test tracks. Driving a car isn't that difficult for a computer system, what's hard is driving in highly complex urban environments with many other cars around that you need to predict.

Planes, in contrast, have a rather simple environment. The number of objects they have to avoid is massively lower, and their freedom of movement is higher, with established rules for how to behave, there are no traffic signs to interpret. This means that all you are really doing is object detection with radar, and collision avoidance.

In addition, modern combat planes are effectively flown by a computer all the time anyway, with the pilot providing the instructions. A number of fighter planes, especially the most modern, are essentially unflyable without computers due to their aerodynamics. Most are inherently unstable around at least one axis, which makes them more manouverable, but means they will not fly stably in the way a 747 will.

The advantage of the "old" control systems approach is that it's not a black box; you can reason about its behavior across the state space and make assertions about its frequency response and stability conditions.

(Is a neural network really different from a very large cascade of nonlinear filter elements?)

> Is a neural network really different from a very large cascade of nonlinear filter elements?

It's different in theory, but not by much practically. Reverse engineering the former is a fool's errand, but so would be a sufficiently complex version of the latter.

In terms of Kahneman's "Thinking Fast and Slow," ML, at least in its current stage, is like the fast thinking system. It's essential but is exponentially more valuable when combined with the slow system, which is still elusive in AI practice.

Indeed. I'm just kind of fascinated by military technology e.g. The F/A-18's cockpit HMDs and Hud were light-years ahead of anything in star wars with the exception of holograms even in 1983.

Coming from a game development background, the easy part is to make the perfect opponent. The hardest part is to let the opponent feel human.

I'm guessing pointing a missile to the right direction has more to do with fast feedback loops and less about training neural networks.

With a missile the actual signal processing and acquisition is what interests me rather than the control theory e.g. you send radar pulses out, you get Doppler shifted pulses back absolutely littered with ECM - now turn that signal into a target to aim at.

Laser codes on A2G missiles are a similar thing that interests me.

> the sheer amount of tech in a modern missile

"Old", reliable tech is prevalent there, too.

Sqlite in missiles is a good example.

Sqlite is used in missiles? Would love to know more about that!

Reply on news.ycombinator.com