Re: Don't forget the orientation
The existing systems on the car had the resolution to do this. The Uber systems that replaced it have better resolution, but crap programming.
6778 publicly visible posts • joined 27 Nov 2009
They are currently only in testing. Until the model is reliable, this is the only acceptable behaviour.
The driver's comfort and convenience in the test phase is irrelevant and they aren't in a hurry to get anywhere, they are being paid to be "driven round in circles". The driver is just sitting there monitoring the vehicle (so they can react if the vehicle makes a mistake), whilst it collects information that can be used to adjust the model to be more accurate.
Safety is of the upmost importance. If the vehicle is in a situation where it isn't sure, it should stop as quickly and safely as possible and let the driver take over.
The movement is key and if the system can't classify whether it is harmless to hit it or not, it should always err on the side of caution and avoid the collision.
If it had said that the pedestrian was a drifting paper bag, that would be one thing. But it just kept throwing away its knowledge that the unknown object was moving to intercept its path and starting from zero with a new static object.
That is the problem. It doesn't track movement, it tracks known objects and if an object changes classification, it has to be dropped and a new one created, because it has different properties. That is an arse backwards way of solving the problem that can only lead to failure.
Until the system has two consecutive sweeps of a particular area, it cannot know if something is moving or not, so the initial status is static. But once it has detected movement, it should use the movement profile as the master and try and classify what is moving, as opposed to reclassifying the object, which resets the motion profile to static.
That is the problem. Either it needs to not drop the object when it is reclassified, but "rename" it, so that the movement profile stays with it, or it needs to use the movement profile as the primary "object" and classify what is moving. But their system just throws away its knowledge of the object and creates a new one.
I expect that they use classes, so you have object, which has child object vehicle, bike, pedestrian etc. and when it reclassifies, it has to drop the vehicle object and create and object object, then drop that and make a bike object, then drop the bike object and make an object object etc.
Better would be to have an object object, which has a "type" classification and loads a movement rules collection based on the type. That way the object and its movement history would remain constant, but the type and its movement rules would be added and removed from the parent object.
That way the movement history and a generic "it is moving to intercept my path" rule would be there even if it can't identify what it is, it can still take avoiding action. Stopping for or swerving around an unidentified piece of trash flying past is silly, but preferable to hitting an unidentified pedestrian.
The processing capability can fit into a car. But you have to do the right processing.
It should detect movement and try and work out what is moving. Instead the system tries to classify what it sees and then work out if it is moving. If it can't classify it or it changes the classification, the movement history is reset to zero and the process starts again.
If the programmers had worked the "natural" way and tracked the movement, whilst trying to classify what is moving, it would have seen much sooner that it was moving across the path of the vehicle, because it has tracked the movement and can predict where it is going, even as it tries to work out WTF is moving.
But the AI was programmed the other way around, it tries to work out what is there first, then tries to give it a movement profile. A re-classification deletes the old object and its movement profile, so you end up with a bunch of static objects that are not in the way.
This is bad programming.
That is the difference between (human) nature and the AI model.
In reality, most creatures recognise a movement and react to that movement, whilst trying to classify what it is as a secondary task. The first reaction is to take a defensive reaction and then work out if it is friendly or not. The pot of petunias.
The Uber AI on the other hand tries to classify the object, before working out if it is moving. The whale.
The problem is, AI programmers have, over the years, worked with object recognition first and movement as a secondary consideration, because it is the "easier" part of programming the model. You work in objects, so you start with the object. If it isn't the object you thought it was, you drop it and start again. This is exactly the opposite of the behaviour that is required. But "movement" isn't an object, it is a metric or property of an object from a traditional programming perspective.
The programmers didn't re-evaluate their logic to fit in with the real world, they just used what they had learnt in programming classes.
The problem is that it didn't know how to classify her, as it wasn't programmed for jaywalkers. Therefore each time she was re-classified, the old object was dropped and a new one created and that new object has no motion history, so is static at the time of capture.
You can see from the log, as the object moves, the AI takes that into consideration, based on its training - a vehicle or bike would usually stay in its lane, so the initial movements show it as moving in its lane.
Then it is re-detected as something else, as it is a new object, the old one, and its movement profile is dropped and a the new object has an empty movement profile. I think that is the problem with the model, instead of re-classifying the object, it deletes the old object and its movement history and creates a new object with no history. If it had recognised the re-classification, it could have worked out that it was moving across its path. This seems like a major failing in the software.
If I see an object moving towards me, it is the movement that is the priority, whilst I try and classify it. The movement continues and continues to be the primary metric I concentrate on, I recognize it as a vehicle, it continues to move and I see it is heading directly towards me. I reclassify it as an SUV, I try to avoid it. It is still moving towards me, I re-classify it not as an SUV, but as a Transit as it gets close and a streetlight illuminates it, but I continue to try and avoid it.
The Uber system works the other way around. Instead of concentrating on the movement and classifying what is moving, it classifies the object first, then looks to see if it moves.
Just because there isn't a law, doesn't mean the discrimination doesn't exist, racism existed long before there were laws to stop it.
But a lot of people seem to be missing that this is geopolitical and going straight to racism, without actually reading the story, or having read the story, have not understood what they have read.
That isn't racial discrimination. That is geopolitical discrimination, big difference.
They aren't stopping Chinese American or Russian American people working for them, they are stopping people living in certain countries that have political differences with GitLab's country of origin from having access to customer data.
I use a private DNS server, which uses DNSSec and DNS over TLS to a major, non-commercial DNS server and my server blocks around 2.5 million tracking and malware sites, including 1,500 Facebook domains.
I had a couple of devices that were bypassing this. I changed my firewall to block DoH and turn on Anycast DNS locally. So far, so good. I get the protection of my local DNS server at home and the lesser protection of DoH on the move.
Yes, it does.
If you develop a new electronic system to process PII, the data protection officer needs to be included from the planning stage onwards. If they are not included in the process (or even you plan the system, build a prototype and then inform the DPO), you are out of compliance.
How many people seeing the Power platform will first talk to the DPO about their plans, before pulling out the company credit card and splurging on a set of Power tools? (Of the Microsoft variety, not the DIY ones.)
I remember running around installing 3com Token Ring cards. *shudder*
A couple of years ago, I was looking at the new offices my employer wanted to move into. (New to them.) It was fully wired, with network ports all over the place, my boss was pleased. I got my tester out and quickly discovered, that all the ports had been wired up for Token Ring and would have to all be re-wired, before we could use the building.
20 years ago? Celebrating 19 years in computing! :-P
40 (well, 39) years ago, playing with a ZX80 and sitting in East Grinstead, England and chatting with the ops in Houton, TX. of an oil exploration firm using VAX Phone. That, plus all the whirring tapes and people walking around with drive spindles to swap out is what hooked me on computing.
And where has instant messaging come to in the intervening 40 years? We now have emojis... Give me the good old days!
exploiting the fact that the antiquated technology behind the UK's remaining pager deployments sends messages without any encryption at all.
So, he is picking up unencrypted messages on the public airwaves? I think it is the NHS Trust that should be in hot water! It isn't too bright of the radio operator to stream the results, but it is clearly the Trust that is at fault here.
It is, however, a criminal offence under both the Wireless Telegraphy Act 2006 and the Snoopers' Charter (aka the Investigatory Powers Act 2016, or IPA) to eavesdrop on messages that are not intended either for the public or for you personally.
As they are unencrypted, it could be argued that the messages are being broadcast in the public domain and he is simply a listener. If they weren't meant to be evesdropped by the public, they would be encrypted.
The webcam side is another matter, of course.
VMWare isn't all Flash based.
The current version uses primarily HTML5 (we have vSphere 6.7.0.30000).
In ESX 6.5 there still wasn't feature parity, some features, such as adding hardware passthru only worked on the Flash client, but with the current version I can add it over the HTML5 UI. I certainly haven't used the Flash version regularly for years - I keep a special VM just for vSphere Flash on our older VMWare servers.
Yes. Replacing a colander with a sieve is always a great move to stop leaks. ;-)
But, yes, before the HTML5 video tag was standardized and supported by most browsers, it was a necessary evil.
I worked at one web agency, where they loved Flash. The whole corporate presence was a single big Flash blob, as was their Intranet. I hated it.
I had to do an internal project in 2008 for product tracking in their photo studio. I wrote it as a website with no Flash, just CSS/HTML5 and, in total, 10 lines of JavaScript on one page from 70 (PHP/MySQL on the backend). They could scan the barcodes using a standard hand scanner, they could do searches, generate reports, all without using Flash or JS.
I got an message on LinkedIn a few weeks back. It was a photographer at the old company who thanked me for the system - 10 years after I left the company! :-O
That is the problem. I have a Pi-Hole, which blocks around 2.5 million tracking, malware and malvertising sites, including all Facebook properties.
But my Fire tablet insisted on using its own DNS server, ignoring my local settings. I had to manually re-configure the settings to force it to use the DNS server it was told to use by DHCP.
The situation will only get worse, when individual applications start ignoring DNS settings.
On public wi-fi or over mobile data, fine. In my house, where I have my own filtered DNS? No!
The only solution is to put in firewall rules to block HTTPS to the Google, CloudFlare etc. DNS servers. But that is beyond most home users.
Yes. If you are running a web server or are running on a cloud, it is an issue.
For businesses who run on-premises or for individual users on their PCs, it is less of an issue.
The proof of concepts have been shown to get the private keys from other instances over a hypervisor. That is a serious risk for multi-tenant servers. But it would only be highly targeted attacks, which means they would be hard to detect and you have to find it, before you know you are being attacked - and if the attack is running on somebody else's instance on the shared server, you may never know, especially if they don't have any AV software or other security in place.
It isn't just external. The GDPR rule is the rule of the minimum. If you don't need to see the data to perform your job, you don't see the data.
For example, sales people have no need to see the contact information of the suppliers and vice-versa and production and support don't need to see either datasets.
Not setting that up properly is a breach. If the company was audited and they found undocumented systems to which employees had unrestricted access to, they would have a big problem.
One of the ground rules of GDPR is that you must involve the Data Protection Officer in all decisions for new systems - and they must be integrated at the planning stage, otherwise you are out of compliance.
If users can install new software and services willy-nilly, without having to go through formal approval services, they are opening a whole bag of hurt for their employer.
We had a lightning strike at our Isle of Wight facility in the late 80s/early 90s. 100s of dead DEC VT100 terminals, but due to the US trade restrictions, we couldn't just dispose of them, these dumb terminals were classified as hi-tech that the Russians shouldn't be able to get their hands on (yes, really, a dumb terminal). So the company and its insurers had to pay for the things to be crushed to a pulp and witnessed by a government official and the certificate supplied to the overseers of the trade embargo.
The proof of the pudding will be when it starts actually returning pertinent results and not shopping recommendations, when you enter a named product.
Searching for "named-product handbook", "named-product error number n" etc. usually returns a dozen shopping results, before pertinent results are returned. If I'm searching for help with a product, I've probably already bought it, or I am in no mood to buy another one!
Even professionals don't get it right every time.
I was in my office in Germany, standing next to the window and stumbled, I reached out and put my hand in the electrical socket - the German sockets have the positive and negative buried out of the way and the earthing prongs stick out of the socket, great for earthing things or clamping the earthing strap to, before working on electronics...
Only the electrician must have been in a hurry and on the one plug where I stuck my hand, he had managed to connect the earth to the positive phase and the positive phase to the earth. I couldn't use my arm for nearly an hour and hurt like hell! The engineers in the production area wouldn't believe me, until they came over with a tester... Followed by quickly turning off the power and righting the electrician's wrongs.
I have to look after some VMs at our subsidiary in America. The VMWare player and Windows don't like the mix of German keyboard on the client and American settings on the host, no matter whether you type the password using German layout of US layout, it won't accept the password, the only way is to use the onscreen keyboard.
I worked at a chemical company. The funniest thing I saw was a computer making random errors. I went to investigate, it was in a shack outside the sulphuric acid store. The fumes at melted away a lot of the contacts on the motherboard!
I told the user we'd have to replace it. He was happy. Until he asked me about copying the data over. I told him the hard drive was corroded as well and it was a good job company policy stated that all data had to be stored on the server... That was when he informed me that the Ethernet port had stopped working 6 months earlier (corrosion from the acid fumes) and he'd stored everything on the local drive. The housing was wafer thin and pitted! I'm amazed I managed to get anything off the drive.
I had a user drop a Klix Hot Chocolate into her Mac keyboard. Luckily she had the presence of mind to yank the cable. The engineer turned up and gave it a thorough cleaning and it worked again. Hardy, those original Mac keyboards on the Mac Plus.
20 years ago? Baby steps. I'm coming up to 39 years in computing. 20 years ago, I was doing OLAP cubes for financial reporting and VB development for a car leasing website, developing the database (SQL Server) and middleware between the database, the IBM mainframe (with MQ Series taking care of the requests) and the ASP pages on the website. It was sold to the client as a 6 week project, I joined the project after 18 months, helping with the writing of the specs and doing coding!
But the add-ons for PowerBI, are they add-ons in the app or are they separate installs? If they are just addons or scripts, they won't need to be installed separately - from reading the article, it sounds like this is buying additional services within PowerBI, so one would assume this is already installed.