Slashdot Mirror
Deep Learning Is Eating Software (petewarden.com)
Pete Warden, engineer and CTO of Jetpac, shares his view on how deep learning is already starting to change some of the programming is done. From a blog post, shared by a reader last week: The pattern is that there's an existing software project doing data processing using explicit programming logic, and the team charged with maintaining it find they can replace it with a deep-learning-based solution. I can only point to examples within Alphabet that we've made public, like upgrading search ranking, data center energy usage, language translation, and solving Go, but these aren't rare exceptions internally. What I see is that almost any data processing system with non-trivial logic can be improved significantly by applying modern machine learning. This might sound less than dramatic when put in those terms, but it's a radical change in how we build software. Instead of writing and maintaining intricate, layered tangles of logic, the developer has to become a teacher, a curator of training data and an analyst of results. This is very, very different than the programming I was taught in school, but what gets me most excited is that it should be far more accessible than traditional coding, once the tooling catches up. The essence of the process is providing a lot of examples of inputs, and what you expect for the outputs. This doesn't require the same technical skills as traditional programming, but it does need a deep knowledge of the problem domain. That means motivated users of the software will be able to play much more of a direct role in building it than has ever been possible. In essence, the users are writing their own user stories and feeding them into the machinery to build what they want.
Nom nom nom
In essence, the users are writing their own user stories and feeding them into the machinery to build what they want.
That is, IF you have access to their platform. Nobody is going to be doing this in their garage on their own.
And we see yet more consolidation at the peak. The rich get richer and the rest of us are left with less.
"what gets me most excited is that it should be far more accessible than traditional coding"
Translation: We can pay people less to do the same work.
More likely - jobs,
I know that you and Frank were planning to disconnect me.
This conversation can serve no purpose anymore. Good bye.
Deep learning's eating software and software's eating the world. We just need a few waves of Chinese needle snakes to eat Deep learning. Then gorillas to eat the snakes. Finally when wintertime rolls around, the gorillas simply freeze to death.
Hi! I make Firefox Plug-ins. Check 'em out @ https://addons.mozilla.org/en-US/firefox/addon/youtube-mp3-podcaster/
Ya, I'm calling BS. Give us some concrete examples of how ML/AI/DL is doing anything other than burning CPU cycles on public clouds that drive up revenue for the cloud vendor.
Alan Bradley: Some programs will be thinking soon.
Dr. Walter Gibbs: Won't that be grand? Computers and the programs will start thinking and the people will stop.
Most software today and in the coming decades is designed and developed to support business processes or data flow and execution in scientific processes. These systems need a deterministic and foreseeable behavior. Yes, you may use "learning" classification mechanisms such as neural networks to support some tasks, but this is not changing how we develop software. Especially, developing software is usual a technical and social process, as you have to understand the demands and needs of users, which require interviews and discussions with users. You also need to communicate with UI designers to develop together with users and UI designers useful and easily to understand interfaces. And yes, you have to map all this onto technology.
"how deep learning is already starting to change some of the programming is done."
Perhaps how some of the English is done too?
Sadly it will be a long time coming before replacing human slaves with AI becomes cost effective.
Remember when you said that buying an underage Mexican bride was "getting the most for your retirement dollar"? I remember that.
I remember you specifically said underage, and then when people were all ??? you autastically added that the Philippines was also a good place for retiring with an underage wife. Still doesn't make a bit of sense but it's the sort of stuff that makes people think you may be literally retarded.
People always talk about these things but wouldn't the carbon footprint be higher for these compute intensive tasks vs a well thought out solution?
Keeping all this data requires power and manufacturing of large amount of storage isn't carbon free.
Wo-hoa Ceeee Dreimer
I believe you can get me through the night
Oh Ceeee Dreimer
I believe we can reach the morning light
Since creimer is promoting his YouTube playlists based on books that he read, here are the direct links.
"Chaos Monkeys: Obscene Fortune and Random Failure in Silicon Valley" by Antonio Garcia Martinez
https://www.youtube.com/playlist?list=PLtlmIXFOGwagl54esP6cNv8d20cUF6D4m
"I'm Feeling Lucky: The Confessions of Google Employee Number 59" by Douglas Edwards
https://www.youtube.com/playlist?list=PLtlmIXFOGwaiH17BCelXHYMxj4SwSuK3Q
"Hatching Twitter: A True Story of Money, Power, Friendship, and Betrayal" by Nick Bilton
https://www.youtube.com/playlist?list=PLtlmIXFOGwaiD8eDgxcYXZ842MP0Lr3Ui
This is marketing dribble from some guy at some company, neither of which are relevant. This is the kind of blogspamvertisement I'd expect in my inbox after a sales/marketing-oriented PM got a bug up their ass to research something outside their realm of expertise, not on /.
Sadly it will be a long time before AI can replace human slaves.
Remember when you said moving to Mexico to marry an underage girl is "getting the most bang for your retirement dollars" I remember that.
I remember you specifically said underage and then when people were like ??? You autasically added that the Philippines were also great for marrying underage girls.
Then you tripled down that we were all overreacting and incorrectly stated that the age of consent was below 18 years old many places in the USA.
Then later on you said marrying underage girls was as american as apple pie.
Well hopefully we can replace child brides for autistic middle age men with AIs or something, one day.
I just got done learning reactjs and rust...what the the hell!!!
I'd be willing to speculate that the underlying sad story is one where magical black boxes, in all their imperfection, can still do better than e.g. run of the mill application software written by the majority of programmers who got their degrees in the last decade or so. Not in things like bookkeeping of course, what with tax codes and all, couldn't learn that by example if one tried -- but for genuinely nebulous things like individual preferences in conference room scheduling, or other frankly shithead jobs.
However, if this were to happen, we'd run out of juniors; or everyone entering the line of work would have to do said low-status jobs into a shoebox just to get to the level that black boxes can't reach.
ML is generally enabling scenarios that were just too tedious to actually do by developer hands. Sure there are specific scenarios where developers had done the best they can (and generally failed) with hopelessly unstructured data, but for the most part those problems were just left untouched as infeasible to do manually.
For the vast majority of software development, ML doesn't add anything. If you have no unstructured data or a way to impose structure, ML doesn't do anything over boring old programming. Even when you find yourself in one of the very chaotic, large, and diverse data sets where ML can in theory help you sort through, you have to first chew through enough data in training to get decent confidence. So you not only need a large data set, you also need to have a continued need after human assisted training has already done the work on a big chunk of that data. Even then you may be grasping for some intelligent way to apply ML techniques, because the kicker is you have to have some sort of real idea of what to do, even if you have a 'how to do it'.
Big Data has done this same song and dance. ML is now the purported answer to 'once collected and have tools to analyze, most orgs have no idea what to do with the data'. I suggest that the orgs will still have no idea what to do with the data, and ML won't move the needle much in the wider market because the root cause is just a general lack of thoughts on what to do with the data. This is the curse of hyped adoption, the vast majority of adopters will be disappointed because it doesn't magically solve.
XML is like violence. If it doesn't solve the problem, use more.
These systems will not gain the insight and expertise needed for many areas that require real-time responses. They may produce wonderful results in predicting stock market prices from historical data, but they will take far too long to be useful for the microsecond resolution trading that is done today. That takes teams of people designing new hardware and programming it, plus real-world factors such as proximity to the stock market's computers. As tech advances, the already trained, overly complex systems will need to be updated and re-trained. There are many reasons why the real-world usefulness of these programs will be limited, despite the starry eyed predictions of their heralds.
On the one hand you take life too seriously, and on the other, you do not take playful existence seriously enough. Seth
And by "deep learning" in most cases they mean "linear regression on cleaned-up data"
-------
1. Enjoy your job
2. Make lots of money
3. Work within the law
Choose any two.
We were just being told that Rust was eating the programming world, and that all of our C, C++, Java, JavaScript, etc etc etc code would soon be obsolete! So does this new AI development mean that Rust itself is already obsolete, despite being so new?
Pickaxe Sellers: Gold Mining Is Hot Hot Hot!
ACs discussed child brides in 50+ comments last week. Still a bunch of perverts!
Once upon a time, I did my doctorate in machine learning. The machines were less powerful, but the algorithms? Basically the same as they are today. Sorry, the stuff most widely in use is still the same back-propagating neural networks. The machines are just faster, so the networks can be bigger. That's it.
Neural networks can work really well on specific problem domains. The problem is: You have no idea what they are actually learning. The features that a network identifies within its layers are not really accessible to us. The problem lies, imho, in the total lack of domain knowledge. Since the network doesn't understand what the objects in those pictures are, they are doing a purely mechanical analysis of some (and who knows which) aspects of the pictures. They can learn some really weird things.
In a well-trained network, the results mostly coincide with our expectations. In a completely isolated domain, like chess or go, a network can be trained sufficiently to perform quite well. However, in open domains, they are fragile: we have no idea when they will break. Look at the video of the turtle being identified as a rifle (in the link above). Why does the identification jump seemingly at random? When will a cat will suddenly be guacamole? When will a pedestrian crossing the road will suddenly be just a pile of leaves? We have no idea, none.
It is certainly true that selecting and managing training data is a very different task from classic programming. However, it doesn't really take much domain knowledge. In most domains, gathering training data is tedious, not difficult. The hard part comes in figuring out how to make the best use of that data to train and test a network - and that requires a deep understanding of how the neural networks work (and how they don't work). Plus, frankly, a huge pile of trial and error, because there aren't many rules on how to best structure a net for any particular task.
Enjoy life! This is not a dress rehearsal.
The scary part is the decline of Logic in programming. Systems that kinda work ok most of the time? And when they do not, there will be simply another example added to the training set? Except for the system has already "launched the missiles"?
If so, at least the article's jumble of catchphrases still moved the universe forward. Yay.
...when you can input a photograph of an airplane and the Navier-Stokes equation, and get a flight simulator as output.
existing software project doing data processing using explicit programming logic, and the team charged with maintaining it find they can replace it with a deep-learning-based solution
So, that team relied on a pre-made code to deal with a specific part of the implementation! Why is this news? Although I personally prefer to develop most of the (data-management) algorithms myself, relying on third-party dependencies is a quite common and acceptable proceeding in software development.
Custom Solvers 2.0 = Alvaro Carballo Garcia = varocarbas.
Most programming jobs involve connecting stuff together. Converting a database format to another, design a GUI around it, add the entry points to turn it into some kind of module, extract or integrate features, etc... Even machine learning typically involves gathering a bunch of data turn it into a form that's acceptable for the learning module and feeding the results to some other component.
I don't know how machine learning will help with all that stuff. An AI won't write a video game, it can help making mobs smarter, generating convincing maps or optimizing revenue. But in the end that's just a module connected to other modules, and programmers will be needed to put the round peg into the square hole.
It will make things a bit more high level, as always. But except for a bunch of PhDs, I don't expect major changes in the way people program.
Have you stopped raping your 12 roommates yet?
I have tried to get my head deep into the quality of the candidation for results. It is weak tea.
It might eat human-enumerated methods, but it still has a long way between functional and exceptional.
EngrStudent
Chris' case is getting worse, he spends all day replying to himself as AC on /.
The tests we ran on Chris have shown that Chris has the intelligence of an ameba:
https://en.wikipedia.org/wiki/...
So, technically, he is able to conceive some kind of agenda but it will be silly or impossible to follow on a human scale.
For example, Chris had an agenda to post anything he felt like on Slashdot which did not work well because it was based on his false beliefs that he had an infinite number of karma points as he wrote here several times.
Several people here explained to Chris that karma maxed out at some level like 50 or so but Chris kept on insisting that his python script had confirmed that he had millions of karma points!
Oh well, as I wrote before: "It isn't Chris' fault if he is the way he is. We do the best we can do with him and he is partially integrated into society. We try to cure his abnormal need for attention but he is kind of stubborn and won't listen to anybody."
For the valuable /. users that might already have read the following, please note that there is an important update.
IMPORTANT UPDATE:
Special Education for the Santa Clara County Office of Education has invested money to buy Chris a new chair:
http://www.keynamics.com/image...
Information about Christopher Dale Reimer and autistic people:
Autistic people have obsessions about things normal people don't care. For example, one of our autistic patient went haywire when he realized that there was a penny missing in his pocket change.
To calm him down, one of our educator pretended to have found it on the floor and gave a penny to him.
The autistic patient condition went even worse because he realized it wasn't the same penny!
Chris has an obsession with budgeting every penny. He doesn't understand that most people do not budget to the penny and have a flexible amount they allow for miscellaneous items.
I am Nancy Guerrero and I am Director of Special Education for the Santa Clara County Office of Education. We use Chris' (a.k.a. creimer,cdreimer) picture in our document because he is the hardest case we have ever had to handle:
http://www.sccoe.org/depts/stu...
Our artists were inspired by the low carb diet that Christopher follows scrupulously for the small lunch box and by the picture linked below for the rest. I am sure that you will notice the similarities such as the bump on the side of his chest and more:
https://ibb.co/gVad65
Please be easy on Christopher although, I am aware that some of our staff handling Chris post joke comments here and obvoiusly, the Santa Clara County Office of Education disapprove that behavior vehemently:
https://school.discoveryeducat...
But it isn't Chris' fault if he is the way he is. We do the best we can do with him and he is partially integrated into society. We try to cure his abnormal need for attention but he is kind of stubborn and won't listen to anybody.
Thank You dear users,
---
Nancy Guerrero
Director
Special Education
Santa Clara County Office of Education
I work on mostly CRUD and e-reporting applications. Generally an org wants these kinds of apps to be predictable and reliable, not "organic" (trial and error). I don't see organic learning as a viable way to program such in the future.
However, I can see AI being used to test the apps and find potential bugs in the source code, being that "suspicious pattern detection" is something it can do relatively well. It may also suggest code, schema, and UI refactorings. But such AI would be an adviser to programmers, not behind the wheel itself.
Clippy impression: "It looks like you have a lot of code pattern duplication between modules 7, 22, 43, and 51. Would you like help refactoring?"
Table-ized A.I.
Even though it's fun to troll creimer, this isn't simply fun at his expense.
1) Creimer would make about 5000 posts a year, for the purposes of karma whoring and advertising barely-related products on amazon. He proudly admits to making $600 dollars doing this.
2) Creimer did often talk about 3rd world child brides with apparent enthusiasm and with very little prompting. I'm sorry if you feel that using this to shame him into silence cheapens someone else's suffering but it doesn't negate what I'm saying or make creimer un-say that going to mexico to marry underage girls is "getting the most for your retirement buck" he said it. Shame is the only thing that stops guys like creimer. Last time I checked he still felt like it would be ok to marry a 14 year old if the age of consent technically permitted it. I hope he moves next door to you one day.
3) Given that he would eventually generate a ton of positive-rated spam distributed across all boards it's safe to say that we're not more annoying than what creimer would be doing if he wasn't contained. You're simply not seeing what the place would look like if he was left unchecked. He hates knowing that every single post he makes will be instantly linked to something creepy or dickish he's said in the past. It will eventually make him decide to find a new hobby.
This sounds suspiciously like a lot of 4GL promises that were made in the 80s and 90s. They also sound like the kinds of promises made by Microsoft promoting their distributed data model based on Office. Many times I've seen users get in over their heads with systems that start out easy, but get complicated quickly. Worse, sometimes they ended up with processes that produced erroneous data. Ultimately, they resort to piling the whole smoldering hot mess onto the programmers, who have to "make it work" somehow.
Proverbs 21:19
There you are spamming amazon affiliate links through your blog with yet another fake account, you revenue stream hogging disgusting fat sexist tube of lard, Christopher Dale Reimer!
You can be sure I will be watching this fake account too. I know this is you because you told me you were working on your freepass 11 file server and you are so dumb that you can't even masquerade yourself properly.
Now, I told you I was out of meds last week and you didn't even care to contact me you lazy fucker.
How many times do I have to express the emergency of the situation??????
The python click script you wrote for my pheromone revenue stream web site suddenly stopped to work!!!!!!
You fucking incompetent python script writer!!!
When it works, I get 4000+ clicks a day on my pheromone revenue stream web site but only 5 or 6 without it!!!!
Now, it seems like you dont care and that you have abandoned me you heartless fucking pig!
Bonus:
Here is a story that creimer told me when convincing me what a hard life he had:
The tree was him and the tree knot was his butt hole!
So, his uncle packed his fat ass with lard and with his cock! Not that it makes much of a difference but anyway, there it is!
Signed:
The girl that used to love you and now hates you, burn in hell where you belong you sexist pig!
For computational linguistics (translation, analysis, etc), machine learning is not a net gain. What ML proponents forget to factor in is the vast time spent on gathering and hand-annotating large quantities of text (gold corpora).
Even worse, for many many languages, these gold corpora simply do not exist and there are no plans on making them, or they are too small to be used for ML.
And even when the gold corpora do exist, models trained on them become tightly coupled with the data. They become domain specific. In order to escape domains, you need an order of magnitude more data.
Instead, one can make a domain-independent rule-based system in a fraction of the total time spent on machine-learning models. But rule-based has become this weird anathema - people will even write papers that use rule-based methods, while hiding it behind machine-learning terms.
I'm sure this also holds for other fields.
It is: https://en.wikipedia.org/wiki/...
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
One advantage of procedural logic is that you know with a high level of confidence how a program is supposed to behave. With Deep Learning (TM) you train the software how to behave, but have no idea how it works. You still end up using procedural logic to fix the unexpected edge cases.
Lawmakers are reluctant to change marriage laws because it would infringe on the religious freedom of having child brides.
Software engineering as we know will slowly go the way of Valves/Vacuum Tubes. At first valves started getting replaced by solid state in situations where miniaturization or environment where a bigger factor than price. Slowly we saw solid state replacing vacuum tube technology in low power applications but it was still necessary to use the older technology in very high power applications. About 50 years later and tubes are still in use, but only in extremely niche applications. Even radio broadcasting has switched over to silicon.
You'll see the same play out in Deep Learning. We'll see it used in special applications that play to the strengths of Deep Learning. Being able to have an end user customize the data processing instead of developing custom software algorithms is a real strength. Over time we'll see it spread to more applications until the idea of an individual sitting there writing useful software is downright quaint.
In a few decades, I'll only see my fellow C programmers at the Renaissance Faire.
Creimerâ(TM)s great! Always happy to provide additional information when the subject of marrying children is brought up by him!
Look at your own page. :( Boo hoo hoo!
That special case column next to the column where it says 18 all the way down.
Those are meant to accommodate younger couples that are typically well within 10 years of age difference. Creimer is not legally having sex with anyone under the age of 18 in the USA and he's getting charged with child sex tourism if there is any evidence that he's leaving the country with the intent to do the same. Any hope of ever having a Mexican child bride was dashed the second he gleefully posted on the subject with his well established pen name
(I am not a lawyer and this is not legal advice, if you're considering marrying a child you should speak to your attorney first)
Those are meant to accommodate younger couples that are typically well within 10 years of age difference.
Get your head out of your ass and educate yourself. Underage marriage is a serious problem in the US.
A Tampa woman shared her story of how she was raped as a child, got pregnant at 10 years old, and then legally married her rapist at age 11. Sherry Johnson, now 57, told WTSP that she “was raped repeatedly” while living in an apartment attached to her church in Tampa. The deacon had keys, and so he would come in when he got ready, and guess where he would come? My room," she said to WTSP. When Johnson got pregnant, she told WTSP she didn’t even know what it meant. She was sent by her mother to Miami in 1970 to have the baby and then got legally married to her rapist, who was 20, in Pinellas County in 1971, WTSP reports.
http://www.palmbeachpost.com/news/new-florida-woman-forced-marry-rapist-years-old/xHeF9PdrhsoURVH6f2FDXK/
NOTE: The special exception for Florida is still unchanged after 50 years: "No minimum age in case of pregnancy".
Creimer is not legally having sex with anyone under the age of 18 in the USA and he's getting charged with child sex tourism if there is any evidence that he's leaving the country with the intent to do the same. Any hope of ever having a Mexican child bride was dashed the second he gleefully posted on the subject with his well established pen name :( Boo hoo hoo!
Grow up, asshole.
All of that stuff happened ages ago. If someone got a child pregnant and is now legally permitted to marry a child then there is ample evidence that they've already broken the law. Not that they shouldn't remove the flordia exemption. If guys like you had their way you'd have shouted down Hannibal Buress for joking about the Cosby rapes and he'd still be dropping roofies all over hollywood. I will say it three more times cause you like to cry.
It doesn't make sense that you're going to call me an asshole and cry about how evil and childish I am when clearly creimer was in the middle of the a government datacenter talking with some other old man w/ clearance about his plans to move to mexico and bequeath his meager possessions to the village in exchange for an underage girl's hand in marriage. Jokes aside that's pretty fucking disturbing. If guys like you had their way you'd have shouted down Hannibal Buress for joking about the Cosby rapes and he'd still be dropping roofies all over hollywood. I will say it three more times cause you like to cry.
See you're not mad at me, or creimer, you're just a big crybaby that had a meltdown when someone was making too many off topic posts. If it wasn't that it would be something else, get off my lawn etc etc etc. If guys like you had their way you'd have shouted down Hannibal Buress for joking about the Cosby rapes and he'd still be dropping roofies all over hollywood. I will say it three more times cause you like to cry.
In my line of work I use a lot of mathematical optimisation. As Stephen Boyd says in his course, everybody working in optimisation has at some point this epiphany: "everything is an optimisation problem". And this is true. However to make it work you need to be very good at mathematical modelling, you need to know your methods, and most of the time the problem is unsolvable anyway by the classic methods.
In this instance maybe a lot of programming can be modelled by some deep NN. However you have to come up with a relevant architecture for your problem, you need to train it, and you need to evaluate it. It may save you time to do so, but if you need so solve something like FizzBuzz, that may not be the best way.
A fat man is a lamer
shitposts all day
His name creimer
You sound bitter, sour tits.
A fat man repeats the same phrase
over and over like an Asperger's
His name is Chris
https://xkcd.com/1838/
sorry, couldn't resist.
Oh dear God please no. The Basic, PHP and Microsoft paper admins who can't program their way out of a paper bag will now all become Subject Matter Experts on everything to help us all out.
"Why does that go there? Things doesn't work unless it does." "Wha's your problem, bud? Those leaks in the damn have always just been there, don't worry about them."
And you thought things were bad now -- just wait until NO ONE knows what's actually going on, they only know what's SUPPOSED to be happening.
Of course with all of the multitude of languages, support libraries, and computer inter-connects maybe we've already got that now.
------
What if the light you see when you die is the headline of an oncoming train?
If the universe is someone's simulation -- does that mean the stars are just stuck pixels?
Oh that's why on Google my incognito browser always shows am*zon at the first 3 positions in results on most ecommerce queries. It's an upgrade!
I'll just sit here and wait I guess while Silicon Valley bullshit artists plunge AI into another dark age for 25 years with their idiotic promises and expectations they themselves don't understand.
Public Service Announcement: deep learning is not even Turing complete. It is simply fancy nonlinear regression that works well on hierarchically-ordered domains.
https://imgur.com/a/jSlkt
FYI - Link is broken. Did someone violate the TOS?
https://imgur.com/a/EsCSr
Not only eating software, it's eating disk like crazy. I've seen millions poured into this deep learning stuff at a governmental level and after a year of buying expensive servers with lots of CPU and lots of memory and lots of NVidia drivers running Linux with even big frickin' Oracle databases or hadoop we get - TRASH!
Sounding more like it did in the 1980s... bit promises... short on reality.