Wikipedia:Articles for deletion/DeepScale

The following discussion is an archived debate of the proposed deletion of the article below. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's
talk page or in a deletion review
). No further edits should be made to this page.

The result was keep. After the third re-list, a consensus formed that it meets GNG/NCORP

(non-admin closure) Britishfinance (talk) 19:48, 13 November 2019 (UTC)[reply

]

DeepScale

[Hide this box] New to Articles for deletion (AfD)? Read these primers!

(Find sources: Google (books · news · scholar · free images · WP refs) · FENS · JSTOR · TWL)

Company lacks in-depth news sources to establish its notability, fails

WP:UPE, who has been editing around the topic. Meeanaya (talk) 07:57, 11 October 2019 (UTC)[reply

]

Thanks for taking the time to review this page. I also feel strongly that Wikipedia should be limited to well-verified content about topics of sufficiently high notability. Whether this page meets the standards is an open question and is worthy of the community's review. I agree that most of the recent coverage is on the topic of the acquisition. That said, the acquisition news drowns out what looks like some fairly in-depth coverage on the technology that DeepScale developed in sensor fusion and deep neural networks that fit on small devices.[1]^[2] I think the main flaw, really, is that EETimes seems to be have been the main source coverage for a number of DeepScale's technology and product updates.

— Preceding unsigned comment added by 104gli (talk • contribs)

References

^ Yoshida, Junko (2017-09-21). "DeepScale on Robo-Car: Fuse Raw Data". EE Times. Retrieved 2018-05-22.
^ Yoshida, Junko (2019-08-25). "Does Your AI Chip Have Its Own DNN?". EE Times. Retrieved 2019-09-26.

Note: This discussion has been included in the list of Companies-related deletion discussions. Coolabahapple (talk) 07:28, 14 October 2019 (UTC)[reply]

Note: This discussion has been included in the list of Technology-related deletion discussions. Coolabahapple (talk) 07:28, 14 October 2019 (UTC)[reply]

Keep. Following the suggestion above, I shortened the history section to focus on the technology, and cleaned up some other less-than-relevant links. I do not think that in the current state it would have been nominated. DGG ( talk ) 17:09, 15 October 2019 (UTC)[reply]
- Hey
  HighKing⁺⁺ 17:41, 18 October 2019 (UTC)[reply
  ]

Relisted to generate a more thorough discussion and clearer consensus.
Please add new comments below this notice. Thanks, North America¹⁰⁰⁰ 12:08, 18 October 2019 (UTC)[reply]

Note: This discussion has been included in the list of California-related deletion discussions. • Gene93k (talk) 02:06, 19 October 2019 (UTC)[reply]

Following up on
User:HighKing's note, here is a reference that appeared just this week on DeepScale's technology. It is not the most high-brow source, but it is not an interview with someone close to the company: https://towardsdatascience.com/what-is-the-technology-behind-deepscale-b40f05fe7423 (apologies for not formatting this reference properly... I am typing on mobile right now). As an aside, there are now over a dozen articles on the acquisition (mostly real journalism, not just reprinting a press release). An other user seemed against these (at the top of this page), but I imagine they would contribute something to the subject's notability. 104gli (talk) 07:00, 19 October 2019 (UTC)[reply
]

There's nothing about the *company* in that article though. The article is discussing the technology only.

HighKing⁺⁺ 13:17, 20 October 2019 (UTC)[reply

]

Thanks for your additional note,

User:HighKing. Could you clarify what other company information you are looking for? Once I hear back from you, I will investigate whether there are references for the types of information that you would like to include. In the mean time, I suggest looking at the following reference. The title of the reference is related to the acquisition, but the reference presents some information on the history of the company, including its people, tech, and product. So far as I can tell, the reference is not based on the journalist interviewing someone from the company. https://www.thedrive.com/tech/30122/tesla-beefs-up-autonomy-effort-with-deepscale-acqui-hire — Preceding unsigned comment added by 104gli (talk • contribs) 16:43, 20 October 2019 (UTC)[reply

]

Hi

HighKing⁺⁺ 18:38, 20 October 2019 (UTC)[reply

]

Relisted to generate a more thorough discussion and clearer consensus.
Please add new comments below this notice. Thanks, Barkeep49 (talk) 05:02, 26 October 2019 (UTC)[reply]

Keep Delete After reviewing references and searching online, I cannot locate sufficient references that meet the criteria for establishing notability. Existing references appear to be based on announcements or research from the company or connected companies, inclusion in "Top 10" type lists, quotations/interviews with people connected to the company and primary sources. Topic fails GNG and
HighKing⁺⁺ 15:49, 26 October 2019 (UTC)[reply
]

Comment. I agree, because While EETimes is a reliable source for the information presented, it is an "electronics industry magazine" which I would argue is a "limited interest" thus is a light weight in consideration of establishing general notability for organization per
WP:AUD. Graywalls (talk) 15:34, 28 October 2019 (UTC)[reply
]

Note: This discussion has been included in the list of Transportation-related deletion discussions. North America¹⁰⁰⁰ 19:30, 26 October 2019 (UTC)[reply]

Keep. This discussion is maturing nicely, and we seem to be getting down to the core of the matter now. As I understand it...

There are good secondary-source references in the following areas: (1) research and technology, and (2) acquisition by Tesla.

In addition, there are sources that discuss the company's products and its corporate partnerships, but these sources appear to rely on interviews with company personnel (making them primary sources per
WP:ORG
.

Here is where HighKing and I seem to have a difference in our interpretation of
WP:ORG that an organization has to be notable for a specific category of thing (say, a product), so long as the organization is notable for something. This organization seems to be notable for its research and technology and for its acquisition by Tesla. This is not uncommon -- for example, my understanding is that OpenAI is also notable primarily for its research and technology. 104gli (talk) 06:48, 27 October 2019 (UTC)[reply
]

HighKing⁺⁺ 20:41, 30 October 2019 (UTC)[reply
]

Delete. Does it have significant coverage? The concept of no-inherited notability is applicable here and the recognition of the acquiring company doesn't weigh i on the notability scale. And "Awards and recognition"s
audience, and all the other ones seem to be mostly about Tesla.Graywalls (talk) 13:17, 4 November 2019 (UTC)[reply
]

Comment.

(1) On no inherited notability, I hear you. Alas, I did not state my point above with sufficient precision. What I meant is that DeepScale's acquisition (irrespective of the acquirer) is something notable, because it was covered by multiple secondary sources, who did independent research, analysis, and fact-checking.

(2) On the awards and recognition, I think
WP:NNC applies here. The awards and recognition aren't being used in the case to establish notability. 104gli (talk) 05:59, 29 October 2019 (UTC)[reply
]
Nonetheless presence of "awards and recognition" in obscure hole in the wall startup is a hint of promotional activity. News outlets regularly announce well established A acquires B, with brief description of the bread crumb "B" being acquired. In this case, the subject of article doesn't have sufficient coverage in depth to satisfy
WP:ORGDEPTH Graywalls (talk) 21:56, 29 October 2019 (UTC)[reply
]

Keep per DGG and 104gli. Passes
WP:SIGCOV. An SNG can not override GNG.4meter4 (talk) 15:58, 3 November 2019 (UTC)[reply
]

The way I understand it from a previous discussion somewhere,
WP:GNG so that business owners and marketing and public relations professionals are discouraged form slipping in pages with promotional interest. Graywalls (talk) 20:03, 3 November 2019 (UTC)[reply
]
Comment
HighKing⁺⁺ 14:17, 8 November 2019 (UTC)[reply
]

I've simply gone through these discussions before and I was pointing out to where you can see
WP:ORGCRIT which says the same thing. This is so you and anyone can see it themselves. Graywalls (talk) 20:49, 3 November 2019 (UTC)[reply
]

@Graywalls: Did you mean to erase my !vote and rationale? Lightburst (talk) 20:06, 3 November 2019 (UTC)[reply]

No. ~~I've simply gone through these discussions before and I was pointing out to where you can see~~
~~WP:ORGCRIT which says the same thing. This is so you and anyone can see it themselves.~~ Graywalls (talk) 20:09, 3 November 2019 (UTC)[reply
]
@Graywalls: Check your edit- My !vote was erased. I guess i will put it back. Please do not erase votes. Lightburst (talk) 20:14, 3 November 2019 (UTC)[reply]
Resolved. There was an edit while I was trying to follow up to 4meter4 while Lightburst was also working on the page and I guess his edit accidentally got cancelled out. Graywalls (talk) 20:45, 3 November 2019 (UTC)[reply]

Keep. per the work of DGG. Passes GNG. Lightburst (talk) 20:00, 3 November 2019 (UTC) Replaced my erased !vote[reply]

Keep per the significant coverage in multiple independent
reliable sources
.

Yushida, Junko (2017-09-21). "DeepScale on Robo-Car: Fuse Raw Data". EE Times. Archived from the original on 2019-11-03. Retrieved 2019-11-03.

Kolodny, Lora (2017-03-21). "DeepScale raises $3 million for perception AI to make self-driving cars safe". TechCrunch. Archived from the original on 2019-11-03. Retrieved 2019-11-03.

Korosec, Kirsten (2019-10-01). "Tesla acquires computer vision startup DeepScale in push toward robotaxis". TechCrunch. Archived from the original on 2019-11-03. Retrieved 2019-11-03.

Kolodny, Lora (2019-10-01). "Tesla is buying computer vision start-up DeepScale in a quest to create truly driverless cars". CNBC. Archived from the original on 2019-11-03. Retrieved 2019-11-03.

Reisenger, Don (2019-10-02). "Why Tesla Quietly Acquired DeepScale, a Machine Learning Startup That's 'Squeezing' A.I." Fortune. Archived from the original on 2019-11-03. Retrieved 2019-11-03.

Lee, Timothy B. (2019-10-02). "Tesla just bought an AI startup to improve Autopilot—here's what it does. We talked to DeepScale CEO Forrest Iandola about his work last year". Ars Technica. Archived from the original on 2019-11-03. Retrieved 2019-11-03.

Cao, Sissi (2019-10-02). "Tesla Just Quietly Acquired a 4-Year-Old Startup to Fill Its Autopilot Talent Gap". The New York Observer. Archived from the original on 2019-11-03. Retrieved 2019-11-03.

Fingas, Jon (2019-10-01). "Tesla reportedly buys AI startup that helps self-driving cars see". Engadget. Archived from the original on 2019-11-03. Retrieved 2019-11-03.

Szymkowski, Sean (2019-10-02). "Tesla reportedly buys machine-learning startup DeepScale for self-driving car tech". CNET. Archived from the original on 2019-11-03. Retrieved 2019-11-03.

Sources with quotes

Yushida, Junko (2017-09-21). "DeepScale on Robo-Car: Fuse Raw Data". EE Times. Archived from the original on 2019-11-03. Retrieved 2019-11-03.
The article notes:
Phil Magney, founder and principal advisor for Vision Systems Intelligence (VSI), called DeepScale’s approach “very contemporary,” representing “the latest thinking in applying AI to automated driving.”
How does the DeepScale approach — using raw data to train the neural network — differ from other sensor-fusion methodologies?
First off, “Today, most sensor fusion applications fuse the object data, not the raw data,” Magney stressed. Further, in most cases, smart sensors produce object data within the sensors, while other sensors send raw data to the main processor — where objects are produced before it is ingested into the fusion engine, he explained. Magney called such an approach “late fusion.”
This is analysis from Vision Systems Intelligence, a technology research company. The article also includes quotes from Forrest Iandola, DeepScale's CEO.

Kolodny, Lora (2017-03-21). "DeepScale raises $3 million for perception AI to make self-driving cars safe". TechCrunch. Archived from the original on 2019-11-03. Retrieved 2019-11-03.
The article notes:
DeepScale’s seed investors included: Bessemer Venture Partners, Greylock, Auto Tech Ventures, Andy Bechtolsheim (who was the first investor in Google) and Jerry Yang. A partner with BVP, Alex Ferrara, said, “Cars are moving from systems today where they have a large number of small computers, called ECUs, in them, to working with smaller more powerful computers for perception. But you have all these little sensors, lidar, radar, ultrasound, and each one brings its own view of the world. There’s a really interesting opportunity here for DeepScale to pull everything together and use info from all those sensors to make computer vision accurate and efficient.”
DeepScale is competing for a share of this burgeoning market versus some 800-lb. gorillas in automotive tech, like Mobileye, now owned by Intel, or Bosch, but also other funded startups like Comma.ai, Argo and Drive.ai, which are trying another approach of building their own, fully autonomous vehicles or retrofit systems.

Korosec, Kirsten (2019-10-01). "Tesla acquires computer vision startup DeepScale in push toward robotaxis". TechCrunch. Archived from the original on 2019-11-03. Retrieved 2019-11-03.
The article notes:
Tesla has acquired DeepScale, a Silicon Valley startup that uses low-wattage processors to power more accurate computer vision, in a bid to improve its Autopilot driver assistance system and deliver on CEO Elon Musk’s vision to turn its electric vehicles into robotaxis.
...
DeepScale has developed a way to use efficient deep neural networks on small, low-cost, automotive-grade sensors and processors to improve the accuracy of perception systems. These perception systems, which use sensors, mapping, planning and control systems to interpret and classify data in real time, are essential to the operation of autonomous vehicles. In short, these systems allow vehicles to understand the world around them.

Kolodny, Lora (2019-10-01). "Tesla is buying computer vision start-up DeepScale in a quest to create truly driverless cars". CNBC. Archived from the original on 2019-11-03. Retrieved 2019-11-03.
The article notes:
Tesla is acquiring DeepScale, a computer vision start-up that could help it develop fully driverless vehicles, CNBC has learned.
The deal could help Tesla’s goal to deliver cars with advanced driver-assistance systems that are good enough for owners to rent them out as “robotaxis” on an Uber-like platform without drivers. However, like all automakers, Tesla is limited by the computational resources it can build into its vehicles.
DeepScale’s technology was designed to help automakers use low-wattage processors, which are standard in most cars, to power very accurate computer vision. These processors work with sensors, mapping, planning and control systems, to allow cars to make sense of what’s going on around them.

Reisenger, Don (2019-10-02). "Why Tesla Quietly Acquired DeepScale, a Machine Learning Startup That's 'Squeezing' A.I." Fortune. Archived from the original on 2019-11-03. Retrieved 2019-11-03.
The article notes:
Now part of the Tesla Autopilot team, DeepScale was previously a Silicon Valley startup with $18.5 million in venture funding that was attempting to develop artificial intelligence technology for fully autonomous self-driving cars.
In order to achieve that, DeepScale relied on deep neural networks, or multi-layered networks that use mathematics and other sophisticated technology to crunch data and deliver real-world information, to create what it called "squeezing A.I.." The "squeezing" means that its technology would use fewer resources to identify obstacles around a vehicle and inform the car's on-board computer to keep the vehicle and its passengers safe.
Artificial intelligence and building fully autonomous driving systems can be expensive. By reducing resource-load, DeepScale's technology could have ultimately reduced costs and allowed more car makers at all levels—from high-line to budget—to implement self-driving technology.
...
Since it's still early days and Tesla hasn't discussed its plans for DeepScale, so difficult to know for sure how DeepScale will find its way into the electric automaker's technology. But there are clues based on what DeepScale was working on and what Iandola posted to his LinkedIn profile.

Lee, Timothy B. (2019-10-02). "Tesla just bought an AI startup to improve Autopilot—here's what it does. We talked to DeepScale CEO Forrest Iandola about his work last year". Ars Technica. Archived from the original on 2019-11-03. Retrieved 2019-11-03.
The article notes:
DeepScale focuses on improving the speed and efficiency of convolutional neural networks, drawing on Iandola's past work as a computer science graduate student. The company's techniques will be particularly helpful to Tesla. Tesla is relying heavily on machine learning techniques to achieve full self-driving capabilities without the lidar sensors or high-definition maps being used by most of Tesla's competitors.
The article includes quotes from DeepScale CEO Forrest Iandola.

Cao, Sissi (2019-10-02). "Tesla Just Quietly Acquired a 4-Year-Old Startup to Fill Its Autopilot Talent Gap". The New York Observer. Archived from the original on 2019-11-03. Retrieved 2019-11-03.
The article notes:
Two sources confirmed to CNBC that Tesla had bought DeepScale “outright,” but were unable to disclose the specific terms.
Until the buyout, DeepScale had raised three venture capital rounds, including a $15 million series A last April led by Steve Cohen’s private investment fund Point72 and Siemens-backed venture fund next47, a $3 million seed round in 2017 and an angel round in 2016.

Fingas, Jon (2019-10-01). "Tesla reportedly buys AI startup that helps self-driving cars see". Engadget. Archived from the original on 2019-11-03. Retrieved 2019-11-03.
The article notes:
An acquisition would make sense. Elon Musk has stressed his belief that Tesla can rely on cameras for autonomy, rather than the bulky LiDAR units many others use. If that's going to happen, Tesla will need self-driving AI that can recognize a wide variety of road objects in less-than-ideal conditions. A buyout like this could bring it one step closer to that reality.

Szymkowski, Sean (2019-10-02). "Tesla reportedly buys machine-learning startup DeepScale for self-driving car tech". CNET. Archived from the original on 2019-11-03. Retrieved 2019-11-03.
The article notes:
CNBC reported Tuesday that Tesla has fully acquired a tech startup company called DeepScale. The startup focuses on computer vision and not on lidar, which many other companies and automakers bank out to give their self-driving car prototypes the gift of sight.
Tesla did not comment on the reported purchase, though CNBC also reported DeepScale CEO Forrest Iandola made a curious announcement on LinkedIn. On the social media network, Iandola confirmed he joined Tesla as a senior staff machine learning scientist. CNBC's sources familiar with the deal reported back saying it wasn't a single hire and that Tesla has, in fact, purchased the startup outright.
DeepScale's approach to autonomy fits the bigger picture Musk has promoted for a few years now. Rather than relying on lidar, Musk has consistently believed cameras, radar and ultrasonic sensors will make up a robust system without other hardware. Powering it all is a new artificial intelligence chip Tesla developed in-house. The chip, detailed this past April, uses minimal power for operation and takes in an absolute massive load of information from the hardware package.

There is sufficient coverage in
reliable sources to allow DeepScale to pass Wikipedia:Notability#General notability guideline, which requires "significant coverage in reliable sources that are independent of the subject".
Cunard (talk) 23:41, 3 November 2019 (UTC)[reply
]

Weak Keep article is written very much from insider's POV and borders on
WP:PROMO. The Tesla acquisition just gets it over the line in terms of notability, but I could see this comping up for deletion at a later point to be merged into the Tesla article.--Goldsztajn (talk) 15:05, 4 November 2019 (UTC)[reply
]

Delete—It appears this company was set up to develop software to the end of being acquired, which it was. I'm feeling this article should be dissolved into the Tesla article and one or more articles on the evolution / development of perceptive systems rather than having a stand alone article. Sorry, folks - I do usually come here to argue for preservation of articles, but in this case, I think that "dissolution" is a better solution = take those citations and ensure that they support content in other appropriate articles. This means the aticle gets deleted, but the content gets incorporated elsewhere. --User:Ceyockey (talk to me) 04:02, 5 November 2019 (UTC)[reply]

Relisted to generate a more thorough discussion and clearer consensus.
Relisting comment: Leaning strongly to a Keep as the Deletes do not seem to be challenging the specific RS being listed by the Keeps (e.g. one-by-one); try a re-list to see if the Deletes can successfully refute them, otherwise, strong lean to Keep.
Please add new comments below this notice. Thanks, Britishfinance (talk) 20:37, 5 November 2019 (UTC)[reply]

Comment An examination of the sources provided by Cunard above. Just to point out that references that are based entirely on company announcements, funding announcements and interviews usually fail the criteria for establishing notability because they do not contain any "Independent Content" which is defined as follows: Independent content, in order to count towards establishing notability, must include original and independent opinion, analysis, investigation, and fact checking that are clearly attributable to a source unaffiliated to the subject. A simple way of thinking about it is that if the journalist does not provide any of their own input in terms of opinion/analysis/investigation/etc and only parrots what either the company itself or others who are linked to the comapny have said, then the article fails. This from eetimes is based on "an exclusive interview" and therefore fails
HighKing⁺⁺ 14:09, 8 November 2019 (UTC)[reply
]

• 104gli's responses are inline below.

HighKing: An examination of the sources provided by Cunard above. Just to point out that references that are based entirely on company announcements, funding announcements and interviews usually fail the criteria for establishing notability because they do not contain any "Independent Content" which is defined as follows: Independent content, in order to count towards establishing notability, must include original and independent opinion, analysis, investigation, and fact checking that are clearly attributable to a source unaffiliated to the subject. A simple way of thinking about it is that if the journalist does not provide any of their own input in terms of opinion/analysis/investigation/etc and only parrots what either the company itself or others who are linked to the company have said, then the article fails. This from eetimes is based on "an exclusive interview" and therefore fails
WP:ORGIND. It's churnalism
.
104gli: When you say, "A simple way of thinking about it is that if the journalist does not provide any of their own input in terms of opinion/analysis/investigation/etc and only parrots what either the company itself or others who are linked to the company have said, then the article fails," it sounds like you are saying is that, if a reporter does an exclusive interview, and then the reporter provides independent analysis as part of the same article, then the entire article is a primary source. However, I think it's often worthwhile to look at an article on a per-paragraph basis to establish what information from the article is primary-sourced, and what information is secondary-sourced. When you click [show] next to Cunard's "Sources with quotes" box, you'll see that Cunard has provided an excerpt of this article. The excerpt is drawing on the thinking of an independent analyst, Phil Magney, who (to my knowledge) is unaffiliated with DeepScale and is quoted in various EE Times articles on autonomous driving that are unrelated to DeepScale (e.g. this and this).

HighKing: This from techcrunch is based on their announcement of raising seed finance and fails
WP:ORGIND
.
104gli: Cunard reproduced an excerpt that I believe is independent analysis: "DeepScale is competing for a share of this burgeoning market versus some 800-lb. gorillas in automotive tech, like Mobileye, now owned by Intel, or Bosch, but also other funded startups like Comma.ai, Argo and Drive.ai, which are trying another approach of building their own, fully autonomous vehicles or retrofit systems."

HighKing: The reference from cnbc, this from Fortune, this from Ars Technica, this from the Observer, this from Engadget and this from cnet are all entirely based on the announcement that the company was being acquired by Tesla.
104gli: A couple of points:

(1) To my knowledge, there wasn't an announcement of the acquisition. It appears that CNBC did some investigative journalism to figure out that the acquisition happened and found sources to confirm the acquisition. If that's not independent investigation, analysis, and fact-checking, I don't know what is.

(2) Take a look at these two passages from Cunard's excerpt of the CNET article: "The startup focuses on computer vision and not on lidar, which many other companies and automakers bank out to give their self-driving car prototypes the gift of sight." and "DeepScale's approach to autonomy fits the bigger picture Musk has promoted for a few years now. Rather than relying on lidar, Musk has consistently believed cameras, radar and ultrasonic sensors will make up a robust system without other hardware." This sounds like independent analysis to me. The journalist is analyzing and synthesizing ideas from multiple sources here, and I think it fits the definition of
WP:SECONDARY. 104gli (talk) 22:41, 9 November 2019 (UTC)[reply
]
Thank you, 104gli (talk · contribs). You have perfectly explained why I provide these quotes: I think it's often worthwhile to look at an article on a per-paragraph basis to establish what information from the article is primary-sourced, and what information is secondary-sourced. When you click [show] next to Cunard's "Sources with quotes" box, you'll see that Cunard has provided an excerpt of this article. The excerpt is drawing on the thinking of an independent analyst, Phil Magney, who (to my knowledge) is unaffiliated with DeepScale and is quoted in various EE Times articles on autonomous driving that are unrelated to DeepScale (e.g. this and this).
Cunard (talk) 00:50, 10 November 2019 (UTC)[reply]
Thanks
HighKing⁺⁺ 13:03, 11 November 2019 (UTC)[reply
]

The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the article's talk page or in a deletion review
). No further edits should be made to this page.

~~Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia:Articles_for_deletion/DeepScale&oldid=926021748"~~

[DataFusion-1] Yoshida, Junko (2017-09-21). "DeepScale on Robo-Car: Fuse Raw Data". EE Times. Retrieved 2018-05-22.

[2] Yoshida, Junko (2019-08-25). "Does Your AI Chip Have Its Own DNN?". EE Times. Retrieved 2019-09-26.

[2]