I don't think YOLO [0], the object detector he talked about, requires a massive amount of data as he claimed. Yes, if you want to learn how to classify 1000 different categories like on ImageNet, then yes, you need a lot of data. But if you're taking a pretrained network like YOLO (it was pretrained on ImageNet and trained on Pascal), you don't need a lot of images. I've retrained it with the KITTI dataset [1] and had no issues at all. They're only 7k images. By the way KITTI actually has a vehicles dataset that might be helpful for your case. And also by the way, you don't even need to retrain YOLO with your vehicle dataset. It was trained on Pascal VOC [2], a dataset of 20 categories and one of the categories is car. So YOLO already knows how to detect cars, it just might not be ideal for your dataset, but you don't care anyways since you just want any solution to compare to as a baseline. This would probably have been even less work than training the cascade classifier you used and have achieved better results.
I feel like all those deep learning papers distorted people's perception of scale. If you need to take those 7k images by hand because your application domain is obscure and they aren't available in an existing dataset, that's way beyond feasible.
You could generate 7k images at a resolution of 4096x2160 pixels by walking around the vehicle for just under four minutes while shooting 4k video at 30 FPS, something of which modern phones are capable.
Yes, but how different would those 7K frames would really be? Same lighting, same background, same surrounding objects, the exact same condition of the vehicle's interior and exterior, same quirks of the camera's color profile, etc, etc. It would be an interesting experiment to actually try this, but I have a feeling the results wouldn't be all that good. Point being, you probably wouldn't get most of the benefits of deep learning and you might as well use the same approach the author used.
No they won't. All of the pictures will have lighting from the time of day and weather conditions from the time and place the pictures were taken. The same problems will happen for the background. If I want my neural network to identify the make and model of cars, but every picture I have of a Mazda3 is taken at noon on a sunny day in suburbia then it is reasonably likely to train on the wrong features and either identify trucks on sunny days in suburbia as Mazda3's or not recognize a Mazda3 photographed on a rainy night.
A human might have difficulty recognizing a Mazda 3 on a rainy night as well. You can adjust color temperature and white balance in post-processing, or film a couple minutes at night too. Point is, generating 7k images is not insurmountable, especially in this case with the criteria that it only has to recognize a particular car.
Reminds me of http://www.nina4airbnb.com/ who made an extensive analysis of where airbnb should target next in an attempt to convince airbnb to hire her.
After her website went viral, she did eventually get an interview with airbnb. While making such an analysis is no guarantee of a job, she was rejected because the interviewer couldn’t contextualize her experience because she "hadn’t worked at facebook or google or studied at stanford".
She did however in the process land several interviews with different companies and picked a company called upwork in the end. So this kind of work can have several positive side effects even if your initial objective is not reached.
According to Nina Mufleh [0] it was the person who had been interviewing her that said that rather than the CMO. Apparently she did not get an interview with the CMO (it was cancelled).
It's exactly the sort of thing people who didn't get a job say was the reason ("ah they didn't like me because I'm not from Harvard") and it's exactly the sort of thing interviewers don't actually say.
Maybe he thought it - he might have said something like "I don't think we're looking for someone with your background" or "we're looking for someone with different skills", but people always read what they think into comments like that.
There's almost no chance he actually said "Sorry we're looking for someone from Harvard" (especially as that is obvious from the CV so you wouldn't even get an interview).
Quite interesting how the authors even mention that this strategy is very economically viable.
From the paper: 'Assuming a selling price of $2 per 1,000 solved captchas, our token harvesting attack could accrue $104 - $110
daily, per host (i.e., IP address). By leveraging proxy services and running multiple attacks in parallel, this amount
could be significantly higher for a single machine.'
The author did compare their performance with captcha-solving services. His accuracy is comparable to the service with no extra cost to the attacker.
From the paper: "We compare our performance to that of Decaptcher, the (self-reported) oldest captcha-solving service. We selected Decaptcher for two reasons. First, it supports the image reCaptcha, charging $2 per 1000 solved captchas. [...] Interestingly, some of our summitted challenges rejected due to the service being overloaded, and had to be resubmitted
at a later time, and received a time-out error as the solvers did not provide an answer in the time window
allocated by the service. 258 challenges (36.85%) were an exact match. When taking into account the flexibility,
321 (44.3%) of the captchas were solved. The average solving time for the challenges that received a solution was
22.5 seconds. While the accuracy may increase over time as the human solvers become more accustomed to the
image reCaptcha, it is evident that our system is a cost-effective alternative. Nonetheless, our completely offline
captcha-breaking system is comparable to a professional solving service in both accuracy and attack duration, with
the added benefit of not incurring any cost on the attacker."
A little hard to take seriously considering the author works for Canonical. Even so the author didn't seem very excited about it, it felt more like an acceptable phone and not that it's better than Android or iOS.
2. The author didn't submit it here, they just put it up on their personal blog. They didn't pretend to be unbiased and then publish a story on a tech news site. They have no responsibility to you just because you found it via a news aggregation site.
3. The author could have easily done this on their own time without any employer involvement at all.
4. Just because the author works for Canonical does not necessarily mean they have any connection to the phone project.
5. The author's employer may have a policy against identifying as a part of the company on personal websites (but no similar policy for social media).
Funny we're often warned we must identify ourselves when publicly praising/reviewing our own products, lest we run foul of the FTC. They are UK based though so perhaps its different there.
I'd go further than "little hard to take seriously".
Nowhere in the article does he state that his employer is Canonical. His blog doesn't mention it at all, just says he's "Involved with Gnome". Even clicking through to his "Complete Profile" takes me to a G+ page. From there another "About" link click finally tells me that he's an employee of Canonical.
Sure, he didn't claim its the 2nd coming of the FSM but stating any potential conflict of interest is pretty standard. A 1 sentence comment on HN from a related employee/founder/ceo usually has a disclaimer about it.
It's a blogpost, made to express an opinion, so he doesn't have to say this. If this was supposed to be a serious article, or was involved in a serious debate, then it would be a different issue.
Second, I have a Firefox phone and no other phone. It's terrible right? Yet I'm happy with it, as it is less buggy than my previous phone, and it actually does what I want. The only thing that I'd like from it would be flac support, but it's something I can get around with a simple script. So, I can see how somebody could be able to live with a Ubuntu phone, and don't need to be paid to say good things about it.
Also, not throwing money at Apple/Google/Microsoft is a a big motivation, and a great pleasure.
Not only is it a personal blog, he did not post the article here. I don't know if he's even aware that it's now on HN. Expectations of disclosures seem absurd.
And at least it seems like Canonical is dogfooding. These wasn't an in-depth review that struck me as masquerading as an objective write up from a non-bias source.
There is no requirement legally or from an accreditation board, sure.
However you should always do this in any context where it may come up. It's not just that this is the ethical path (it is) but failing to do so and getting caught out really undermines your credibility and thereby whatever message you were trying to get across (serious or no). There is literally no downside to doing this unless you are trying to obfuscate your connection.
Nonsense. "View my complete profile" takes you to the G+ page, and directly underneath his name it says he works for canonical. So what if his personal blog doesn't disclaimer everything in it - the information was one click away. You're just looking for excuses, really.
This new HN where everyone is complaining of 'clickbait' titles and demanding disclaimers everywhere for things like personal blog posts... it's starting to get pretty tiresome.
> Nonsense. "View my complete profile" takes you to the G+ page, and directly underneath his name it says he works for canonical. So what if his personal blog doesn't disclaimer everything in it - the information was one click away.
A further click on "About" in the header, shows that he is an employee of Canonical.
> You're just looking for excuses, really.
Actually I was looking for confirmation or denial of his previously unstated employment by Canonical, the company whose OS runs on the device he reviewed without making mention of his conflict of interest.
> This new HN where everyone is complaining of 'clickbait' titles and demanding disclaimers everywhere for things like personal blog posts... it's starting to get pretty tiresome.
This is nothing to do with HN. He posted a review of a product that his employer is directly related to, and made no mention of who he works for. Most people I know would consider it fishy to not make any mention at all of your involvement with a product you're reviewing.
just to clear up a misunderstanding between you and vacri you're responding to: Google seems to be in the process of yet another redesign of G+.
If you haven't yet enabled the new experience then the employer is stated quite prominently directly on that link. This is what vacri was referring to.
If you enable the new experience, then yes, you are right, you have to click one more time.
Yes. I'm splitting hairs here - this should probably have been on the blog-post front and center, but still. You two are not seeing the same content.
I wonder how much faster the new Tesla P100 is compared to the Tesla K40 in training neural networks. The K40s were the best available GPUs for training deep neural networks.
I thought Jeff Dean was the 'Chuck Norris' of programming. He designed (with others) a lot of Google's biggest infrastructure projects as well as multiple iterations of the main search infrastructure.
Among others, the projects he's worked on include:
Spanner - a scalable, multi-version, globally distributed, and synchronously replicated database
Some of the production system design and statistical machine translation system for Google Translate.
BigTable, a large-scale semi-structured storage system.
MapReduce a system for large-scale data processing applications.
Google Brain a system for large-scale artificial neural networks
LevelDB an open source on-disk key-value store.
TensorFlow an open source machine learning software library.
That's what I was thinking when I saw him on the list. I've been studying top industrial, FOSS, and CompSci projects for years. I see certain names, like Dean or esp DJB, that turn up over and over in high-quality, clever work they help code. Then, I find they instead list someone who was answering questions on SO. Hmmm.
To be fair, in case he deserves the credit, I did ask him to elaborate in another comment. Most programming in our industry is people solving day to day problems. They encounter all kinds of issues that hold our collective productivity back. Having a one-stop answer to most is very valuable. I think it's worth giving Skeet a high honor if he's done a huge chunk of that and done it well. I just think it should be a different title or type of honor along the lines of educating programmers instead of one that seems to imply his skill of designing/programming.
If Jeff Dean built Google, Jon Skeet is my Google.
Atleast until Google is able to tell me where my program is supposed to store local files on the twenty different versions of Windows.
I have a big respect for Jon Skeet, but to dismiss Jeff Dean's work because Skeet is the one who provides copy-and-paste snippets for programmers is wrong on so many levels.
I'm not an european citizen, have a job offer to work in a big tech company in Zurich and will not be able to get a visa even though I've got my Bachelors and Masters degree in Europe. Since the referendum passed, they're now requiring 3 years full-time work experience regardless of which education degree you hold.
Well, they're voting for their own Swiss issues, not even european (as in "EU european") issues. Political correctness aside, I don't think they're questionable.
I highly doubt this has to do with cover letters. From Google's Job FAQ:
"Cover letters are optional, and applications with them are not preferred over those without. We recommend using a cover letter only to bring up anything that would not be evident from the rest of your application. It can also be an opportunity to let Google know about you as an individual, in addition to why you're a good fit for the organization. We strongly recommend using your resume to represent your candidacy."
http://www.google.com/intl/cn/jobs/students/sga/mba/internsh...
From a Google Screener:
"As a heavy interviewer and occasional resume screener at Google, I rarely look at the cover letter. Your resume should give me enough context for whether you have the key experience for the role."
http://www.quora.com/How-should-a-cover-letter-for-Google-lo...
Right. Based on one of the other comments I've gone to have a look at that. Things have changed, and may be significantly different at places like Google as opposed to my experience. I've only ever been hiring at startups and small companies, and never in the US.
I guess it's also different for an internship as opposed to a job. As someone with 20 years of experience in running companies, when faced with a bare CV I can never tell what someone will be like in a company. I always like to see someone tell me why they add value, and then back up their claims with evidence from their CV, but I guess that doesn't apply here.
All that being said, if all they look at is the CV then a bare recitation might not be the best presentation. It would be interesting to look at how to make a better "CV" for this purpose.
Anyway, thanks for the information - appreciate it.
I've went through the same thing as OP. I'm a student located out of the US that applied for internships at all major companies in the US (Microsoft, Facebook, Google, Apple, Palantir, etc) for summer last year. I really have to disagree with people mentioning cover letters or costs to bring international students. I didn't submit any cover letter and got accepted for two of those companies for an internship in the US and accepted an internship with one of them.
Google for example really doesn't care about cover letters. This used to be at Google's Job FAQ:
"Cover letters are optional, and applications with them are not preferred over those without. We recommend using a cover letter only to bring up anything that would not be evident from the rest of your application. It can also be an opportunity to let Google know about you as an individual, in addition to why you're a good fit for the organization. We strongly recommend using your resume to represent your candidacy."
http://www.google.com/intl/cn/jobs/students/sga/mba/internsh...
By a Google Screener:
"As a heavy interviewer and occasional resume screener at Google, I rarely look at the cover letter. Your resume should give me enough context for whether you have the key experience for the role."
http://www.quora.com/How-should-a-cover-letter-for-Google-lo...
I've noticed you've probably studied a lot already on how a resume look like, I can see you're using Gayle's template for CVs and I did the same thing. Your CV is pretty good and better than mine, it should definitely be attractive to recruiters because of your awards (ACM ICPC 2015 World Finalist) and your previous experience (multiple internships).
Here's the thing, the hard part is getting a recruiter to actually read your CV. Coincidentally or not, only the companies to which I was referred by full timers actually bothered to interview me. I have the impression the strength of the referral also makes a huge difference. For one of the companies my referral was from another intern and I didn't get an interview. Was your referral from an FTE or another intern?
The one difference I can pinpoint between us is that you're applying from the middle east and I applied from Europe. I'm really not sure how much of a difference this makes, but I can tell you that I did not see many foreigners from that region during my internship last year. I'm not sure this should actually make a difference, because my visa situation wasn't quite simple either and the company didn't seem to have any issue with the fact I'm from south america and applied for a J1 visa while living abroad in Europe. My guess would be your CV isn't getting read and to try to get stronger referrals.
Makes a big difference. Applying for work visas for someone from the EU is totally different than applying for a work visa for someone from the middle east.
I'm not an EU citizen. I doubt top tech companies would reject a candidate just because it's more visa work. They usually have an entire immigration department just for handling these cases.
You have no idea what a pain the US visa process is. For instance, for regular employees your employer needs to obtain a H1B visa. There is a maximum of 85,000 a year, which is reached in one week for the last 3 years.
[0]: http://pjreddie.com/darknet/yolo/
[1]: http://www.cvlibs.net/datasets/kitti/
[2]: http://host.robots.ox.ac.uk/pascal/VOC/