Particle physicists turn to AI to cope with CERN’s collision deluge

https://www.nature.com/articles/d41586-018-05084-2

by okket • 8 years ago

132 61 8 years ago

nature.com

JumpCrisscross2• 8 years ago

Out of curiosity, what fraction of the Bitcoin network’s computing power would solve this computation wall?

amelius• 8 years ago

Is there even a bound on the computational power we'd need to make an advancement here?

JumpCrisscross2• 8 years ago

Although these techniques would be able to work out the paths even after the upgrades, “the problem is, they are too slow”, says Cécile Germain, a computer scientist at the University of Paris South in Orsay. Without major investment in new detector technologies, LHC physicists estimate that the collision rates will exceed the current capabilities by at least a factor of 10.

Sounds like the LHC’s physicists have an idea.

jrq• 8 years ago

No, but CERN Is known to be careful with their PR. Presumably, and this is just my intuition speaking, a big enough cluster of computers would solve this, but they're taking an opportunity to experiment with different techniques and methods for this experiment, and that's pretty much it.

If CERN had an unlimited budget, I suspect they'd do it however they did it before.

21• 8 years ago

Zero. All the Bitcoin network can do is compute SHA-256 hashes. It's literally useless for anything else.

zamalek• 8 years ago

I think they might mean the raw computational power (excluding ASICs).

SiempreViernes• 8 years ago

Dunno, how much memory and storage does the network have?

The CERN grid currently[1] has 1 EB of storage and 750k of CPUs, and that pushes out 2 million jobs per day. From own experience of one of the 162 sites, you have something like 6 GB of ram per CPU core, and often jobs need more than that so that in practice you are memory bound.

[1] https://indico.cern.ch/event/466934/contributions/2524828/at...

jamesblonde• 8 years ago

They also have 0 GPUs. (I visited there a couple of weeks ago and talked to them, among other things, about this challenge).

tim333• 8 years ago

Maybe they could issue Cerncoins, one issued for each particle traced. Then if they went up in value the miners could switch to that.

dean177• 8 years ago

“The top three performers of this phase hosted by Google-owned company Kaggle, will receive cash prizes of US$12,000, $8,000 and $5,000.”

“incomparably more difficult”

lima• 8 years ago

It's a competition, people aren't doing it for the prizes.

mkagenius2• 8 years ago

Yes, but many certainly are doing it for the prizes.

mkagenius2• 8 years ago

Its very economical for companies to have 100s of AI programmers work on this (for a few of months) for $25k.

But surprisingly the value they get out of it is certainly only 1 team's work.

itissid• 8 years ago

Can't they ask the top K teams for their solutions. Offer some smaller prize to them. I believe Kaggle supports that. Even if you have 10 top teams of 5 people each and gave them a sum of 300K total its damn cheap than hiring 4 or 5 full time engineers working for < 1 year.

jononor4• 8 years ago

Usually in Kaggle they ask every team elegible for a prize for their solution. I believe it is just as much to ensure there was no cheating/hacking, than to actually make use of the solution itself.

jononor4• 8 years ago

This competition (and the followup) is official for important conferences in the field, so I expect many teams to publish their results. In general there is lots of discussion, tips & tricks posted online during such competitons, both in Kaggle forums and in blogs. So the value is certainly more than that off the winner, and most of it is available to everyone, not just the sponsors.

konschubert5• 8 years ago

What's the news here? Using Machine Learning in data analysis is common practice in particle physics.

Is the news that they want to use ML in the trigger selection now?

Also, which one of the four experiments is doing this?

EDIT: Ah, it's CMS.

diehunde2• 8 years ago

I think it's that they posted part of their data on Kaggle to see if someone can come up with a better model

dukwon6• 8 years ago

Is the news that they want to use ML in the trigger selection now?

Not quite (and that wouldn't be news). It's using ML for track reconstruction. Not even LHCb does this.

konschubert5• 8 years ago

Ah, cool! That's really quite interesting!

tempay• 8 years ago

LHCb does use ML for trigger selections already[1].

[1] https://cds.cern.ch/record/2243560?ln=en

ibeckermayer• 8 years ago

I was working in UBC's ATLAS lab last summer, and this appeared to be the main focus of nearly all the PhD student's research. Using ML models for track reconstruction.

konschubert5• 8 years ago

The problem with AI in particle physics is that you need to very well understand the efficiency of the selection ( rate of false negative and false positive classification). And this rate of course isn't uniform for all kinds of events. Thus, the uncertainty on the selection efficiency tends to grow with the complexity of the machine learning model. This directly hurts your sensitivy, the very thing you were trying to improve by using an AI - based process.

It's a trade-off.

jessriedel• 8 years ago

From what I remember doing this years ago, the selection efficiency is usually determined through numerical simulation rather than analytically. So why would your estimate of the efficiency get worse with more complicated models?

dukwon6• 8 years ago

In general, simulation doesn't agree with data perfectly. The more the variables (and their correlations) aren't reproduced faithfully, the more the efficiency determined from simulation is likely to depart from the truth. It's even possible to accidentally train an algortihm to discriminate between simulation and data rather than signal and background.

It's common practice to check data/simulation agreements between variables before using them for training. Reweighting procedures are used to improve agreement.

Analog244• 8 years ago

As already pointed out, detector efficiency is determined through extremely detailed simulations of the entire system. Those measurements are done in a completely orthogonal manner so the complexity of the model has nothing to do with the systematic uncertainties that arise from the selection efficiency.

konschubert5• 8 years ago

As already pointed out, detector efficiency is determined through extremely detailed simulations of the entire system.

It depends on the experiment, LHCb for example does not use simulated background.

In any case, the more complex your model gets (number of variables) the exponentially more simulated Monte Carlo events you need to fill that multidimensional space.

jononor4• 8 years ago

Curious to see what kind of techniques end up doing best on this. Worked on using GPUs in this areas back in 2010 for the ALICE project. Which was pretty straight port of an existing algorithm for finding certain phenomena given tracks. I believe the track reconstruction then used a multistage process ending with a Kalman filter.

nonbel• 8 years ago

"In the new problem, she says, you have to find in the 100,000 points something like 10,000 arcs of ellipse."

Does anyone know how long this currently takes?

danbruc6• 8 years ago

That certainly depends on the amount of computing power you can throw at it and what level of accuracy you are trying to achieve. But the most naive attempt using a Hough transformation will probably take a couple of milliseconds on your average computer. However, the accuracy will be far from good enough with such a simple approach.

I have no idea what amount of time state of the art algorithms will use, but I could well imagine that it is essentially a tunable parameter that is chosen to get the best accuracy given the amount of data you have to process and the time available to complete the task.

mattheww5• 8 years ago

Hough transform is basically useless for precision tracking in a high multiplicity environment. Tracks have 5 degrees of freedom, so the memory costs make it infeasible. I think ATLAS uses it in the trigger, where you don't need to actually reconstruct all tracks, just find out if there are a couple passing certain criteria.

mattheww5• 8 years ago

Depends exactly how you limit the problem, but going from detector hits to abstract arcs takes a few seconds.

amenod• 8 years ago

Someone posted what looks like a very nice summary of domain knowledge:

https://www.kaggle.com/pranav84/beginner-s-guide-to-cern-s-p...

jononor4• 8 years ago

Here is a very to-the-point summary of particle tracking: http://www.physics.iitm.ac.in/~sercehep2013/track2_Gagan_Moh... - link found in Kaggle forums

danbruc6• 8 years ago

A simplified model of the task. You throw 10,000 springs of various sizes (particle trajectories) into a box and record the intersection points (hits) with a set of (spring-penetrable) nested cylinders (detectors) in that box. You will get on average about 10 intersection points per spring. Now given those 100,000 intersection points, reconstruct the size, position, and orientation of the 10,000 springs. Lets add the facts that all the springs are aligned along one axis and that the cylinders are nested around that axis to be geometrically closer to reality.

The hardness is due to possible ambiguities of what collection of springs caused the observed intersection points, imperfections in the helical shape of the springs and the cylindrical shape of the detectors, limited resolution of the measured intersection points, and missed (detector efficiency »only« 99 %) or added (detector noise) intersection points.

Coding_Cat• 8 years ago

Plus, the detector is multi-layered around the point of collision. Each layer measuring something slightly differently with different accuracies. Some springs are even completely invisible and you can only find them by looking at what is missing (neutrally charged particles).

mattheww5• 8 years ago

The main thing of interest here is whether a system that knows about physics (traditional approaches) can be beaten by a system that has no a-priori knowledge of physics (OOB DL) or if someone will find a way to integrate physics knowledge into a DL approach.

DrNuke• 8 years ago

Hard hard problem, good luck to competitors. The new Kaggle-Google is a cheapskate trap but still fun a couple hours a week reading the forums and the most brilliant kernels.

jaddood• 8 years ago

I am sure the people at CERN have reasons in choosing machine learning for that purpose, but I don't see it ideal for this case. Machine learning might be able to give you your results faster, but generally with a lot lower accuracy. But suppose the teams got to a really good accuracy in their programs, it's still for the known and expected. In particle physics research — or any research for that matter — you cannot simply give it a shot, because your system might miss very interesting collisions while you think it's working perfectly fine. I think quantum computing can be really quite useful here given that the analysis can benefit from parallel processing. Should ML be necessary, I recommend develop it in a way that its doings are understandable. (I'm not sure they'll even read the comment to get the recommendation, but let's give it a try.) Explainable AI is becoming really significant these days, and some big companies and organisations are working on it. If I'm not mistaken, the DARPA is working on a project as such, so it's really not too far off.

Particle physicists turn to AI to cope with CERN’s collision deluge

Related Stories