Tag Archives: machine-learning

16 January 2020
3 Comments
Bez kategorii

DeepL – a translator which surpassed Google Translate

Reading Time: 4 minutes

A company doesn’t have to be a technological giant to create a product that exceeds the most popular programs of the same type. There is no doubt that in the world of automatic translation Google, Microsoft, and Facebook are the leaders. And yet it turns out that a small company DeepL has created a translator that sometimes exceeds the quality of the most popular programs of this type.

DeepL logo
Source: https://www.deepl.com/home

 

How DeepL was created?

It turns out that the key to the development of the translation service was to use the first own product, which is Linguee, a translation search engine on the Internet. The data obtained in this way became training material for artificial intelligence behind DeepL.

Interestingly, Linguee’s co-founder, Gereon Frahling, once worked for Google Research but left in 2007 to continue his new venture.

Currently, DeepL supports 42 language combinations between Polish, English, German, French, Spanish, Italian and Dutch. Already now, artificial intelligence is learning more, such as Mandarin, Japanese and Russian. There are plans to introduce an API, by means of which it will be possible to develop new products and implement the mechanism in other services.

The team has been working with machine learning for years, for tasks bordering on basic translation, but finally, they began a fervent work on a completely new system and a company called DeepL.

 

What is the advantage of DeepL?

Once again, people realized that AI is learning all the time – to the benefit of consumers, of course. The artificial intelligence behind the DeepL not only accurately recognizes words and selects translations, but is also able to understand certain linguistic nuances, perfectly copes with changed sentence patterns, which makes the result of a user’s inquiry extremely natural – as if it was written by a human being.

The company also has its own supercomputer, which is located in Iceland and operates at 5.1 petaflops. According to press releases with such equipment DeepL is ranked 23rd in the Top 500 supercomputers worldwide.

 

The statistics do not lie

The blind test compared the new product and solutions from Google, Facebook, and Microsoft. Professional translators were supposed to choose the best results of the mechanisms in the comparison without knowing the author of the translations:

DeepL’s blind testing results
Source: https://techcrunch.com/2017/08/29/deepl-schools-other-online-translators-with-clever-machine-learning/

 

But that’s not all, because in the BLEU results DeepL also gets great scores. BLEU is an algorithm for evaluating the quality of translation.

 

Why do others recommend DeepL instead of Google Translate?

The main advantage of DeepL in the context of Google Translate is much better knowledge (or rather a detection) of idioms, phrases, and phraseological compounds. Where, for example, Google Translate is weakening and literal meaning is being found, DeepL can surprisingly offer a more nuanced and much more specific language solution. The translation is not a literal translation of the text, but one that best harmonizes with the contexts and connotations characteristic of the words.

The passage from a German news article rendered by DeepL
Source: https://techcrunch.com/2017/08/29/deepl-schools-other-online-translators-with-clever-machine-learning/

The passage from a German news article rendered by Google Translate
Source: https://techcrunch.com/2017/08/29/deepl-schools-other-online-translators-with-clever-machine-learning/

 

No wonder that DeepL is gaining recognition all over the world. Here are some reviews:

Thanks to more French-sounding phrases DeepL has also surpassed other services. – Le Monde, France

In the first test, from English to Italian, it was very accurate. In particular, he understood the meaning of the sentence well, instead of being stunned by the literal translation. – La Repubblica, Italy

DeepL from Germany surpasses Google Translate. A short WIRED test shows that the results of DeepL are by no means worse than those of its best competitors, and in many cases even surpass them. Translated texts are often much more fluid; where Google Translate creates completely meaningless word strings, DeepL can at least guess the connection. – WIRED.de, Germany

We were impressed with how artificial intelligence selects the translations and how the results of its work look afterward. Personally, I had the impression that on the other side sits a man who on speed translates. – Antyweb, Poland

 

The DeepL tool has been made available to a wider audience – for free in the form of a website.

Now it is only a matter of waiting for DeepL to advertise its tool, because although it does not have a large language base, at first glance the accuracy of the translations definitely exceeds the most popular tools of this type.

It’s worth watching how the product will develop further as the current achievements of DeepL are really promising.

Did any of you choose DeepL instead of Google Translate?

 

References:

[1] https://techcrunch.com/2017/08/29/deepl-schools-other-online-translators-with-clever-machine-learning/

[2] https://www.deepl.com/blog/20180305.html

[3] https://www.dw.com/en/deepl-cologne-based-startup-outperforms-google-translate/a-46581948

[4] https://www.forbes.com/sites/samanthabaker1/2019/06/27/will-this-german-startup-win-the-translation-game/

[5] https://www.deutsche-startups.de/2018/07/05/deepl-koelner-uebersetzungskoenig-macht-millionengewinn/

[6] https://www.forbesdach.com/artikel/davids-erbe-und-igels-strategie.html

[7] https://www.letemps.ch/societe/deepl-meilleur-traducteur-automatique

Tagged , , , , , , ,
6 January 2020
2 Comments
Bez kategorii

Duolingo – the best machine learning startup to work for in 2020

Reading Time: 3 minutes

Duolingo is starting the year off strong. They have been named one of the top startups to work for, in the growing field of machine learning. These and many other insights are from a Crunchbase Pro analysis completed using Glassdoor data to rank the best machine learning startups to work for in 2020. Why is Duolingo a unique company?

Duolingo logo
Source: https://www.duolingo.com/

Duolingo AI Research

Duolingo AI Research is one of Duolingo’s fastest-growing teams. They are using real-world data to develop new hypotheses about language and learning, test them empirically, and ship products based on their research. Duolingo has revolutionized language learning for more than 300 million people around the world. They keep on bringing creative, interdisciplinary ideas on how to deliver a high-quality education to anyone, anywhere, through AI.

 

Duolingo AI team logo
Source: https://research.duolingo.com/

 

Tools and data from Duolingo

Duolingo use AI to adapt longer learning content to learners’ level. The startup is regularly releasing their internal tools to the public so everyone can read more about their research innovations. One of them is CEFR Checker. This tool determines whether texts are appropriate for beginner, intermediate, or advanced learners of English or Spanish. It works by analyzing vocabulary and highlighting words by their reading proficiency level according to the Common European Framework of Reference (CEFR). Duolingo uses interactive tools like this one to help people revise content (e.g., Podcasts and Stories) for particular levels.

The Duolingo CEFR Checker: an AI tool for adapting learning content
Source: https://making.duolingo.com/the-duolingo-cefr-checker-an-ai-tool-for-adapting-learning-content

The Duolingo CEFR Checker: an AI tool for adapting learning content
Source: https://making.duolingo.com/the-duolingo-cefr-checker-an-ai-tool-for-adapting-learning-content

 

Duolingo is also committed to sharing data and findings with the broader research community. SLAM Shared Task is an example project. It contains data for the 2018 Shared Task on Second Language Acquisition Modeling (SLAM). This corpus contains 7 million words produced by learners of English, Spanish, and French. It includes user demographics, morph-syntactic metadata, response times, and longitudinal errors for 6k+ users over 30 days.

 

Why people should consider working at Duolingo?

The language-learning app Duolingo is valued at $1.5 billion after a $30 million investment by Alphabet’s CapitalG. Bookings growth has risen from $1 million to $100 million in less than three years for the most downloaded and top-grossing education app worldwide. What is more, Pittsburgh’s first venture capital-funded $1 billion start-up plans to increase staff by 50% with the new funding. Duolingo has been adding user and revenue at an impressive pace, continuing to solidify its position as the No. 1 way to learn a language globally.

 

Why people should consider working in the machine learning field?

Demand reminds high for technical professionals with machine learning expertise. According to Indeed, Machine Learning Engineer job openings grew 344% between 2015 to 2018 and have an average base salary of $146,085 according to their Best Jobs In The U.S. Study.

It can be safely stated that Duolingo is developing very dynamically. There is also no doubt that the rapid growth of a startup also means the development of its employees.

Would you choose to join Pittsburgh’s unicorn if you had such a chance? What do you think about Duolingo’s contribution to the development of the education sector?

 

References:

[1] https://www.forbes.com/sites/louiscolumbus/2020/12/29/the-best-machine-learning-startups-to-work-for-in-2020-based-on-glassdoor/#71505e744886

[2] http://blog.indeed.com/2019/03/14/best-jobs-2019/

[3] https://www.cnbc.com/2019/12/03/google-funded-duolingo-first-1-billion-start-up-from-pittsburgh.html

[4] https://making.duolingo.com/the-duolingo-cefr-checker-an-ai-tool-for-adapting-learning-content

[5] https://making.duolingo.com/how-machine-learning-helps-duolingo-prioritize-course-improvements

[6] https://cefr.duolingo.com/

[7] https://research.duolingo.com/

Tagged , , , , ,
1 January 2020
1 Comment
Bez kategorii

Spotify’s Tastebuds tool will enhance your social music discovery

Reading Time: 3 minutes

The new function aimed at Spotify will allow us to slightly broaden our musical tastes. All thanks to our friends from the platform.

New Spotify’s feature logo
Source: https://techcrunch.com/2019/12/18/spotify-tastebuds/

 

How music sharing options look like now?

Thinking about what Spotify can offer today, we can mention social media integration. It is a popular feature that enables users to connect their Spotify accounts to their Facebook and Twitter profiles. That lets them access their friends’ favorite music and playlists and share their choices with others as well. Individuals can create, share, and edit playlists with other listeners. If users want recommendations, they can integrate their system with Last.fm, an application that provides music recommendations based on listening history. However, this is an external application that does not belong to Spotify. You can also view a Friend Activity ticker of songs your Facebook friends are currently listening to on the desktop app. You can search for specific users and follow them or view playlists they’ve made public too. Spotify doesn’t promote user search much anyway.

 

Why Spotify need a change?

Sharing playlists on Spotify is not a problem, but when it comes to speeding up the entire algorithm, things get complicated quickly. If we listen to music in a random way, it is known that Spotify will not start sending us recommendations based on it. If that were the case, playlists created using algorithms would not suit certain user’s tastes. Social sharing has never been the main priority for Spotify. The Activity Feed, which shows what your friends are currently listening to, is limited to the desktop version of Spotify. The in-app messenger for sharing music was nixed in favor of letting users share songs on social media or on their Instagram Stories. Apparently, that was a mistake as far as we know that Tastebuds is coming.

Tastebuds feature on Spotify
Source: https://techcrunch.com/2019/12/18/spotify-tastebuds/

 

But what will Tastebuds really give us?

As the information on the site is telling: Tastebuds will let your friends discover music that you trust. This description appears in the tab that has not yet been launched, but the developers have already sewn it in the application – in the left column, next to the Home Page, Browse and Radio. The prototype feature was discovered in the web version of Spotify by reverse engineering sorceress Jane Manchun Wong. She gave some more details on how it works. By tapping on the pen icon, users can view information about what their friends have been playing most. Then, they can easily listen along or add songs to their own library.

Tastebuds feature discovered in the web version of Spotify thanks to reverse engineering
Source: https://techcrunch.com/2019/12/18/spotify-tastebuds/

Spotify Tastebuds code
Source: https://techcrunch.com/2019/12/18/spotify-tastebuds/

 

It remains to wait for deployment

When will the new feature come into effect? There is no official information about it. A Spotify spokesperson confirms that they are always testing new products and experiences, but have no further news to share with the audience. For now, anyone can access a non-functioning landing page for the feature at https://open.spotify.com/tastebuds. Tastebuds could be a rebranded version of the Friends Weekly playlist that was discovered in May 2019. Whatever it may be, the test could be a sign of more social listening features to come.

Social is a huge but under-tapped opportunity for Spotify. Social recommendations could get users listening to Spotify for longer. While competitors like Apple Music or YouTube might offer similar music catalogs, users won’t stray from Spotify if they become addicted to social discovery through Tastebuds.

Do you think Tastebuds is just what Spotify needs? Maybe something else would make the application more user-friendly? Share your opinion.

 

References:

[1] https://www.crunchbase.com/organization/spotify#section-overview

[2] https://rms.pl/aktualnosci/sprzet/3239-spotify-tastebuds

[3] https://www.spidersweb.pl/2019/12/spotify-tasetbuds-czego-sluchaja-znajomi.html

[4] https://www.theverge.com/2019/12/18/21028474/spotify-tastebuds-playlist-friends-music-discovery-social-sharing-feature

[5] https://techcrunch.com/2019/12/18/spotify-tastebuds/

[6] https://www.engadget.com/2019/12/19/spotify-social-music-discovery-tastebuds/

[7] https://www.theverge.com/2018/5/9/17337182/spotify-testing-new-friends-weekly-playlist

Tagged , , , , , , ,
19 January 2019
1 Comment
Bez kategorii

The use of facial recognition technology on birds

Reading Time: 2 minutes

Today, I want to demonstrate you a great example of how object recognition technologies based on machine learning:

1) becoming widely available and do not require rare genius programming skills to get the result.

2) can be greatly trained even on a very modest in size data sets.

The article, that I have read some time ago, tells how a bird lover and part-time computer science professor, together with his students, taught the neural network to recognize the bird species and then — and that impressed me a lot – to distinguish individual species of woodpeckers, who flew to the bird feeder in his yard.

At the same time, 2450 photos in the training sample were enough to recognize eight different woodpeckers. The professor estimated the cost of a homemade station for the recognition and identification of birds at about $ 500. This is really can be called technology for everyone and machine intelligence in every yard.

Moreover, this technology can really help birds. As Lewis Barnett, the inventor of this technology wrote in his article : «Ornithologists need accurate data on how bird populations change over time. Since many species are very specific in their habitat needs when it comes to breeding, wintering and migration, fine-grained data could be useful for thinking about the effects of a changing landscape. Data on individual species like downy woodpeckers could then be matched with other information, such as land use maps, weather patterns, human population growth and so forth, to better understand the abundance of a local species over time»

As some people correctly noted, this technology has also some great commercial potential. Just imagine that camera traps will be able to recognize birds that harm your fruit trees and than activate  a device that make a large noise to scare away pests.

Sources:

https://theconversation.com/i-used-facial-recognition-technology-on-birds-106589

Tagged , , ,
2 December 2018
Leave a comment
Bez kategorii

How Artificial Intelligence is starting to have a serious effect on our lives?

Reading Time: 4 minutes

Have you seen Minority Report directed by Steven Spielberg? 

 

 

For those who haven’t I recommend watching it because the prophecy of this film begins to meet.

Due to the fact that technological process is constantly developing and thanks to that the meaning of Artificial Intelligence in our lives increase, we can definitely be scared about this what is happening around us.

 

Have you ever been thinking about which is one of the most intimate things in people lives?

It is sexuality orientation. Nowadays many people hide them real sexuality in fear of social indignation for example: sport players, family members, schoolmates. This people have to bother with this inside battle of “coming-out” every single day and now it is going to be worse. Nowadays the AI can guess whether you are gay or straight based of photos of your face. It is the fact not the opinion! Now we can say that machines started to be better “gaydar” than people. The study work from Stanford University – has found algorithm which could distinguish with 81% of accuracy whether you are gay or straight for men and with 74% of accuracy for women.

The algorithm was tested on machine intelligence which had to research of 35 000 facial photos from the one of dating sites and thanks to that had find out the real sexual orientation.

“The research found that gay men and women tended to have “gender-atypical” features, expressions and “grooming styles”, essentially meaning gay men appeared more feminine and vice versa. The data also identified certain trends, including that gay men had narrower jaws, longer noses and larger foreheads than straight men, and that gay women had larger jaws and smaller foreheads compared to straight women.”   – Sam Levin, The Guardian

 

Okay what if I am straight?

The authors of study which was published in the Journal of Personality and Social Psychology, Dr. Michal Kosinski and Yilun Wang, claim that this algorithm can also be used as a similar AI system which could be trained to spot others human traits such as IQ or political views. They are also warning us against this AI develop process because it can turn into something that we don’t really want to in our lives.

It is happening now!

Police in the UK are piloting a new project which provides to use AI to determines how someone is likely to commit the crime. Seems familiar? Back to that what I wrote at the beginning of my post, Steven Spielberg (Director) and Philip K. Dick (writer) were right. AI is going to prevent us from committing the crime.

 

“(…) The system has 1,400 indicators from this data that can help flag someone who may commit a crime, such as how many times someone has committed a crime with assistance as well as how many people in their network have committed crimes. People in the database who are flagged by the system’s algorithm as being prone to violent acts will get a “risk score,” New Scientist reported, which signals their chances of committing a serious crime in the future. (…)

(…) Donnelly told the New Scientist that they don’t plan to arrest anyone before they’ve committed a crime, but that they want to provide to those who the system indicates might need it. He also noted that there have been cuts to police funding recently, so something like NDAS (National Data Analytics Solution) could help streamline and prioritize the process of determining who in their databases most needs intervention. (…)”
– Melanie Ehrenkranz, gizmodo.com

The project now is in its infancy in comparison to how important it can be for the future of the justice system.

To sum up my post, there are billions of facial images of people that are publicly available on social media sites, government databases and also these ones which come from the streets cameras. In my opinion we should try to care more about our privacy in a media and don’t let the governments to have that serious impact on our lives because as we know the systems are like people, they sometimes fail.

 

 

 

Sources:

https://bit.ly/2Pg2urN
https://bit.ly/2BMzAMk
https://bit.ly/2AHTgiE
https://bit.ly/2AGC9Og
https://bit.ly/2So735a

Photos:

https://bit.ly/2KONqAT
https://bit.ly/2BL0jJ1
https://bit.ly/2AHX9Eg
https://bit.ly/2riurpi

author: Michał Żelazo

Tagged , , , , , , , , , , , , , , , , , , , , , ,
2 December 2018
Leave a comment
Bez kategorii

No more f-word while driving

Reading Time: 2 minutes

How often a poor road surface drives you to madness? It would be great to visit car mechanic only on the occasion of a routine control. Wouldn’t it?

Inventors of the US company called RoadBotics found an answer to this problem. They offer a smartphone-enabled road inspection tool that creates the map of the condition of the road surface. The solution based on deep learning algorithms examines data from a video made by smartphone located behind the windshield of the vehicle. It recognizes problems such as cracks and potholes. In this way, the tool analyzes a 10-foot road section and assigns its level of damage. The results are presented on a five-point scale where 1 – green is an excellent surface condition and 5 – red is failed.

“This assessment does not suggest what needs to be done, but serves as a triage tool, allowing your engineers and public works staff to narrow down on what requires immediate attention. It complements your current process and enables your municipality to reduce time spent on inspecting roads and focus on maintenance.”[1]

Savannah is the city that has already employed this solution. RoadBotics has analyzed half of the streets and roadways in the city within a few months at the costs of nearly 25 000 $. Normally, the city would have to hire interns who would take 3 years to perform an analysis of the road conditions in the entire city.

Heath Lloyd, the City’s Chief Infrastructure and Development Officer said:

“This partnership with RoadBotics will allow us to better assess the overall condition of City roadways and increase the accuracy of the data collected. It ultimately allows us to be more effective in managing the replacement of our infrastructure”[2]

RoadBotics was introduced to URBAN-X, the leading accelerator for startups. It is constantly working on developing its product, among others using it in self-driving cars.

 

 

Sources :

https://techcrunch.com/sponsored/this-company-created-a-high-tech-solution-to-a-2500-year-old-problem/

https://www.urban-x.com/company/roadbotics/

https://www.savannahga.gov/CivicAlerts.aspx?AID=1927

 

 

Tagged , , ,
25 November 2018
4 Comments
Bez kategorii

The new, best way to deal with cyber-attacks?

Reading Time: 4 minutes

Darktrace autonomous AI-system finds out about any digital threats, before they get severe.

Information leaks happen almost every day in the Internet. Meanwhile most of the engineers try to find a way to prevent hackers from getting into the digital systems, some of them noticed that it might be better to do it the other way around. In 2013 with that idea in mind, Darktrace has been created.

A group of former MI5 agents joined forces with Cambridge mathematicians with a mission in mind to develop a new tool to fight cyber-attacks. Interestingly enough, they decided to use AI to make that happen. As their philosophy states ‘Pit the machines against the machines to keep your data safe.‘.

So how does it really work and why is it effective? Frankly, it is quite simple. Darktrace connects their software with company’s system. From that moment, the AI starts monitoring all the activities that occur within the digital infrastructure. Furthermore, it is learning how does the company operate.

Well, you may now ask yourself a question why is it all for? With all of the data accumulated, Dartktrace’s software can now easily detect any instances of unusual activities or deviations. This is called unsupervised learning, a very rare type of machine learning, that doesn’t require any information from us humans, to know what to look for.  This really revolutionized the market, because before that we used supervised learning, which is quite the opposite. In that case we had to provide data to the AI in order to allow it to learn about the threats and problems that may occur. Although it works fine in most of the cases, it has its flaws too. The main problem is that it is useless when unknown threats appear. That’s where Darktrace has the advantage.

Darktrace software here just neutralized an anomalous, dangerous behaviour

For example, in 2017 the software was introduced to one of the Las Vegas casinos. Although the company states, that their AI usually is not really useful within first days of working due to its learning process taking at least a week, just after its start, it registered some unusual activities. It turned out, that their recently installed fish tank, which had electronic sensors connected to the servicing company, had transferred over 10GB of data to an external device, which did not belong to the company. After some digging they have found the hacker all the way in Finland.

As Dawn Song, a cybersecurity and machine-learning expert at the University of California, Berkeley stated “the whole system is as secure as its weakest link” and that is the great example of that.

An example of how Darktrace interface looks like.

What also accumulates to their superiority over the market is the accessibility. Their software is really easy to use and to see through. They also provide consultations, if anyone from the IT department encounters any problems with the software. Although co-chief executive Poppy Gustafsson said that they do not want to focus on that part of service “We don’t do consulting” she said “Our tech is not just about detecting cyber threat but also to autonomously respond.”.

Also interesting is the fact that the whole idea was inspired by human body. In one of the interviews, the co-CEO of the company, Nicole Eagan said “It’s very much like the human body’s own immune system,” and moreover  “As complex as it is, it has this innate sense of what’s self and not self. And when it finds something that doesn’t belong—that’s not self—it has an extremely precise and rapid response.”

This start-up has been performing incredibly ever since it was created. In March 2015 they were evaluated at 80M dollars. Only three years later in September 2018 they are valued at over 1,65 billion dollars. This rapid growth is was mainly accelerated by Mike Lynch and his Venture Fund, Invoke Capital. He owns right now over 40% of the company making him the shareholder with the highest ownership.

Although right now it may seem for you, as if this is a perfect software and solution to cyber-crime, it has its flaws too. Some IT workers had reported that this AI-based system, continuously reports multiple deviations throughout the day, to the point when they had to stop checking the alerts, just because it was a waste of their time. Furthermore, Darktrace’s plans for their customer are not cheap at all, which can make them less desirable.

Frankly, I would say that even though it will help bigger companies to eliminate some threats, especially from the inside, it is nowhere near the perfect solution yet.

What do you think about this start-up? Are AI-based systems the solution to our problem with cyber-crime? Let me know in the comments.

 

 

Reference list:

  1. Leslie, I. (2018, June 15). You used to build a wall to keep them out, but now hackers are destroying you from the inside.
    https://www.wired.co.uk/article/darktrace-insider-threats-hackers-security
  2. Ram, A. (2018, October 10). Inside Darktrace, the UK’s $1.65bn cyber security start-up.
    https://www.ft.com/content/2fa5bade-cb09-11e8-9fe5-24ad351828ab
  3. null. (null). Cyber-Security SEIM | IDS. https://msp-partner.com/darktrace/
  4. Clifford, A. (2018, August 7). How billion-dollar start-up Darktrace is fighting cybercrime with A.I. .
    https://www.cnbc.com/2018/08/07/billion-dollar-start-up-darktrace-is-fighting-cybercrime-with-ai.html
  5. Hao, K. (2018, November 16). The rare form of machine learning that can spot hackers who have already broken in.
    https://www.technologyreview.com/s/612427/the-rare-form-of-machine-learning-that-can-spot-hackers-who-have-already-broken-in/
  6. Darktrace. (null). Company Overview. https://www.darktrace.com/en/overview/

 

Tagged , , ,