felippee

10 ms·

31.
by felippee 8y ago
I'll happily read your next post where you will include all of those. In fact amount of VC money spent in that field would only support my claim. And the number of papers is irrelevant. There were thousands of papers about Hopfield net
32.
by felippee 8y ago
> The difference between the current AI renaissance and the past pre-winter AI ecosystems is the level of economic gain realized by the technology I would argue this is well discounted by level of investment made against the future. I do
33.
by felippee 8y ago
Thanks for that, that is essentially my point. Agree it is not very rigorous, but it gets the idea across. By scalable we'd typically think "you throw more gpu's at it and it works better by some measure". Deep learning
34.
by felippee 8y ago
Author here: I'm using deep learning daily so I have a bit of an idea on what I'm talking about. 1) Not my point. Hype is doing very well. But narrative begins to crack, actually indicative of a burst... 2) DL does not scale very
35.
by felippee 8y ago
Yeah, the problem is deep learning sucked a bunch of money - essentially took a loan against the future in the form of VC investments. And if that loan does not get payed, for the next few years you may not afford to explore all that other
36.
by felippee 8y ago
Hi, author here: Well first off: letters to investors are among the most biased pieces of writing in existence. Second: I'm not saying connectionism did not succeed in many areas! I'm a connectionist by heart! I love connectionism
37.
by felippee 8y ago
Author here, yeah, it is the autumn. But I guess not many people would recognize the meaning, winter on the other hand is not ambiguous...
38.
by felippee 9y ago
Yeah depends on what you mean by close. I see the cheapest 1080ti at newegg right now for $909. Bought mine a year ago (founders edition) for $699. It is getting better but we are still far away from sanity. IMHO this (year old) card should
39.
by felippee 9y ago
The number you pull out is not telling. There are billions of miles driven in the US alone and people get killed on average after 100 mil miles driven. That is where the bar is, and I can tell you, that is a high bar. Even including all the
40.
by felippee 9y ago
Tesla was not a self driving car, please stop confusing this. This case is in fact the first autonomous test vehicle caused fatality. There were at least several Tesla autopilot related fatalities and injuries, but I would seriously not pu
41.
by felippee 9y ago
Though I agree with the spirit of what you are saying, I would add that it is perhaps not the weakness of neural networks per se, but weakness of the current architectures and training (supervised, hence verbalized) paradigms. I think we co
42.
Some new ideas for AI benchmarks in vision(blog.piekniewski.info)
3 points by felippee 9y ago|0 comments
43.
by felippee 9y ago
It is 0.99 deaths/100 mil in California, but fair enough. I'm not sure where you got your "well over 10mil miles" data. Last time I checked, Waymo clocked 3mil, Uber 2mil, Cruise <0.5mil, and the rest are small potato
44.
by felippee 9y ago
I'm sorry to disappoint you, but again the data does not support your claim. The cumulative number of miles driven by autonomous test fleet to date is nowhere near 100 million miles, which is where one should expect the first fatality
45.
by felippee 9y ago
LOL, all I can say to this, is that perhaps you should read your Tesla manual. I sincerely wish you luck with your driving, just watch out for fire trucks parked in the emergency lane...
46.
by felippee 9y ago
What self driving data do you mean? There are no self driving cars in deployment right now. There are various semi-autonomous safety features that _REQUIRE_ full attention from the driver.
47.
by felippee 9y ago
You are mixing things. Measuring safety of a human driven car with a human driven car with additional safety feature is not the same as measuring safety of autonomous vehicle. Contrary to its name, autopilot is not autonomous driving featur
48.
by felippee 9y ago
This is a huge misconception. Modern AI is at least 30 years old, first convnets were conceived in 1980'ies, LSTM in early 90's. A lot of work went into deep learning before AlexNet eventually won the ImageNet. To some degree last
49.
by felippee 9y ago
You should probably reconsider your trust in the face of data: https://blog.piekniewski.info/2018/02/09/a-v-safety-2018-upd... But who cares about data, if it is all about AI singularity religion?
50.
by felippee 9y ago
An alternative view on the presented data can be found here: https://blog.piekniewski.info/2018/02/09/a-v-safety-2018-upd... Including telling plots comparing various safety numbers, along with all the data s
51.
Latest California disengagement data – Autonomous Vehicles are still far away(blog.piekniewski.info)
3 points by felippee 9y ago|1 comments
52.
by felippee 9y ago
This is not nitpicking. This is actually quite important. The entire AI hype rides on such irresponsible analogies. These analogies completely obscure the actual state of the art and fuel some outrageous expectations (not to mention the sin
53.
by felippee 9y ago
This is great work. It shows that some of the "amazing" results of deep learning are not as deep and don't even require learning! In my view this also sheds some light on the GAN's and their ability to generate "rea
54.
Nouvelle AI 2.0 – An alternative direction for AI research(blog.piekniewski.info)
3 points by felippee 9y ago|0 comments
55.
by felippee 9y ago
Let me explain: motherboard has 1000 components. One of them fails and it is likely a brick. SSD contains several chips. Most probably what will fail will not be an SSD but some capacitor somewhere on CPU power line. With a separate SSD, yo
56.
by felippee 9y ago
Yeah sure. But sometimes you are traveling, without access to your backup. Or something did not yet make it to the backup. Or whatever, 100 other situations. Making SSD soldered on as opposed to putting a $3 connector just adds to fragility
57.
Artificial Intelligence Committee – Hearing in the UK Parliament on Impact of AI(parliamentlive.tv)
4 points by felippee 9y ago|1 comments
58.
by felippee 9y ago
It's going to be a one hell of a software update...
59.
by felippee 9y ago
Well this is exactly my point. A lot of problems are currently framed as computation problem. As much as it might be useful for some, I'd argue that we should be careful with this. We get all the great marvels of deep learning, but rob
60.
by felippee 9y ago
I think the major problem is to think of "intelligence" as a problem in "computation". We are so used to this framing, that it may seem foreign. Others will argue, that in principle it is a problem of computability but i

More ›