The successful verification of Huawei Cloud Pangu’s large model was completed under the rigorous testing of the China Academy of Information and Communications Technology. The model’s ability to support language, speech, vision, and multi-modal tasks has been comprehensively evaluated, and it performs particularly well on key indicators such as data compliance, model controllability, and service reliability. This not only proves the technological advancement of Huawei Cloud Pangu large model, but also reflects its ability to meet the high standards of the financial industry.

The financial industry has extremely strict requirements for AI technology, involving the accuracy and security of data processing as well as the continuity and stability of services. The outstanding performance of Huawei Cloud Pangu large model in these aspects provides a solid foundation for its application in the financial industry. For example, in terms of data classification and attack prevention, the model can effectively identify and process sensitive information to ensure the security of financial data; in terms of output accuracy, its high-accuracy results help financial institutions make more accurate decisions. decision-making; in terms of local deployment, scalability, and compatibility, the Huawei Cloud Pangu large model shows strong adaptability and flexibility, and can meet the personalized needs of different financial institutions.

The success of this verification is also another important milestone for Huawei Cloud in promoting the development of the artificial intelligence industry. Looking back on the past, Huawei Cloud’s layout in the AI ​​field is far-reaching and solid. From infrastructure construction to algorithm research and development, from platform services to industry solutions, Huawei Cloud continues to promote technological innovation, actively participates in the formulation of industry standards, and explores the broad applications of artificial intelligence with partners from all walks of life.

At the Second Beijing Artificial Intelligence Industry Innovation and Development Conference held on February 29, Huawei and 23 other enterprises and institutions jointly launched the “Large Model Application Industry Consortium”. The consortium aims to rely on Huawei’s Kunpeng hard base and soft base to jointly explore the artificial intelligence ecological cooperation model, establish artificial intelligence industry standards, create an artificial intelligence industry ecosystem, and accelerate the upgrading of the artificial intelligence industry. This move will not only further promote Huawei Cloud’s in-depth development in the AI ​​field, but will also provide new impetus for innovation and progress in the entire industry.

Looking to the future, as artificial intelligence technology continues to mature and application fields continue to expand, the successful verification of Huawei Cloud Pangu’s large model will bring more possibilities to the financial industry. From intelligent customer service to risk management, from investment analysis to credit approval, Huawei Cloud’s AI technology will help financial institutions improve efficiency, reduce costs, and enhance competitiveness. At the same time, this achievement of Huawei Cloud will also inspire the entire industry to continue to promote technological innovation and jointly embrace the bright future brought by artificial intelligence.

To sum up, the Huawei Cloud Pangu large model received the highest rating in the financial AI standard compliance verification, which not only affirms Huawei Cloud’s technological innovation capabilities, but also actively promotes the intelligent transformation of the entire financial industry. This achievement of Huawei Cloud will bring a more efficient, safer, and smarter service experience to the financial industry, and also set a new benchmark for the innovative development of the artificial intelligence industry.

The Peng Cheng Cloud Brain II, a supercomputer developed by Huawei and Peng Cheng Laboratory (PCL) that carries software and hardware with artificial intelligence (AI), broke a record and won two world wide-scale entry and exit championships.

According to the latest IO500 ranking from the International Supercomputing Conference (ISC21), the input and output power of this computer’s overall system is nearly 20 times that of the runner-up.

The supercomputer is capable of collaborating in artificial intelligence research and exploration, such as computer vision, natural language, autonomous driving, intelligent transportation, and smart healthcare.

It features a Huawei Atlas 900 artificial intelligence cluster powered by Huawei Kunpeng and Ascend 910 processors. The Atlas 900 completes training of a ResNet image classification model in 59.8 seconds, 10 seconds faster than the previous world record with the same precision.

The Atlas 900’s powerful computational capabilities offer a difference when it comes to collaborating on scientific research and technological innovation, as well as astronomical exploration, weather forecasting, autonomous driving and even oil exploration.

This is AI, this is technology, this is art, this is love, this is the Arcfox Alpha S, the first Huawei’s electric car! At the Shanghai Motor Show, The giant tech group Huawei presented its electric car, the Arcfox Alpha S. The Chinese brand got ahead of other technology groups by announcing on April 17, this model developed with the help of the group, Beijing Automobile Industry Co (BAIC).

Extensive use of new technologies
The Arcfox Alpha S will be equipped with a Kirin 990A chip, also used in other Huawei products, and in particular some of the brand’s smartphones. Kirin 990A uses 4 Taishan V120 cores (small) + 4 cores based on ARM V7A series. The Vortex A55GPU part uses Mali-G76. The chip also includes Huawei’s own Da Vinci architecture computing power microchips. The presence of the Kirin 990A chip aims to develop artificial intelligence in the car and allow a connection to 5G thanks to the modem integrated into the chip.

This first model will offer a level 3 autonomous driving system. This level is the first of the six levels of autonomy which considers that the car can only depend on itself to operate. The vehicle will support three levels of autonomous driving including the NCA mode which takes into account, thanks to a preinstalled card, various points in order to achieve automatic driving. Other driver assistance technologies are also present in the vehicle.

Compete against Tesla in the future
Among the other technologies that can be found in the car, we can note a 12-speaker system or the presence of a 21-inch 4K touch screen, based on HarmonyOS . Thanks to this platform developed by Huawei and Android, this touch screen will have everything of an integrated smartphone. A multitude of navigation and entertainment applications will be offered.

In order for it to function, the Arcfox Alpha S will be equipped with a 93.6 kWh lithium-ion battery allowing it to display up to 708 kilometers of autonomy. It will be accompanied by a fast charger to recover 195km of range in ten minutes.

Currently, the arrival of Huawei on the electric car market is also correlated with the Chinese firm’s desire to dethrone Tesla, which is already very present on the market.

Do you want to use Machine Learning and AI in your company or for a project, but you do not have the expertise of a Data Scientist or an engineer? Check out the top 10 best tools for building machine learning and artificial intelligence applications without computer code.

Artificial intelligence and machine learning are opening up new possibilities for companies in all industries. However, these technologies require the expertise of Data Scientists and other specialists with programming knowledge.

However, on a global scale, there is a real shortage of such experts. Fortunately, to facilitate the adoption of AI in business, several tech giants offer open source platforms and tools to exploit this innovation without expertise in computer code. Here is a selection of ten such tools.

Teachable Machine
The Teachable Machine is a web-based tool . From a simple browser, the user can create machine learning models accessible to everyone. It suffices to feed the computer with examples to enable it to learn. These examples can be files, or live captures. Once downloaded, the examples are categorized into images or audio.
The models can then be tested immediately to see if the new samples were correctly classified. Thus, it is possible to teach models to classify images, sounds or even body postures.

What-If Tool
The What-If Tool is distinguished above all by its ease of use . User can easily run two models on the same dataset to compare differences through visualization features.

The data points can be edited at will by adding or removing features. Finally, it is possible to perform a test before putting the model into production.
Another strong point of this tool is the use of confusion matrices and ROC curves to determine the accuracy of models.

Google AI Platform
The Google AI Platform is Google’s artificial intelligence platform . It stands out for its affordable price and ease of use.

With this solution, Data Scientists can easily bring their ideas to life through an integrated toolchain to run a machine learning application .

This platform is compatible with Kubeflow, Google’s open source platform . This allows the user to design portable pipelines that can be run on Google Cloud or on premise.

The data is first stored in the Cloud or on BigQuery , then labeled to classify it in different categories: images, videos, audio, text, etc.

The data can then be imported to train a model . A machine learning application can then be created on the Google Cloud Platform (GCP). This supports different machine learning frameworks using a virtual machine deep learning image. The AI ​​platform and the GCP Console make it easy to manage models.
Data Robot
With Data Robots, you can create AI projects in minutes instead of months. This solution allows you to exploit predictive models without needing the expertise of a Data Scientist.

The tool provides access to different open source machine learning models . You will thus be able to obtain models whose accuracy will depend on the data available.

Thus, a company is able to take advantage of predictive models without needing to develop its own set of proprietary models. In addition, this platform offers a balance between machine learning and human experience to solve predictive modeling problems.
RapidMiner Studio
The RapidMiner Studio offers you a very intuitive “drag and drop” interface . After collecting all the available data, the platform will choose from a collection of over 1,500 algorithms to determine the best model.

The company’s databases, data warehouses and social networks are quickly connected to allow the user to share the data with anyone who needs it .

When the model is ready, the tool explains why it is the best choice and what benefits it will bring to the business. In addition, visualizations make it easy to explain the workflow to anyone inside or outside the company.

Accelerite ShareInsights by Amazon Web Services
Powered by AWS, ShareInsights is a tool for designing ETL (extract, transform, load) pipelines of data without the need for computer code.
A drag and drop interface makes it easy to create the pipeline by relying on the various Amazon cloud services . Cloud-native technologies like Glue and Arena can be leveraged to create interactive dashboards in minutes.

A data analysis platform for S3 and Redshift is also offered, as well as automated service selection and hit management features for AWS serverless services. There are also data preparation, OLAP and Machine Learning functions.

Create ML By Apple
Create ML is a very easy to use application. It allows the user to deploy Machine Learning models , without the need for technical knowledge in Machine Learning.

The user can view the model building workflow in real time . He can also develop his own models for object detection, activity or sound classification.

Many models can be trained using different sets of data simultaneously. Models can then be tested before being deployed. The application is designed to work without a dedicated server. It is possible to improve performance by using an external GPU .

Microsoft Azure Automated Machine Learning
The Microsoft Azure cloud-based automated machine learning tool enables organizations to deploy machine learning models much faster .

Its user interface does not require coding , and the tool automatically deploys predictive models using existing data filtered by algorithms and hyperparameters.

Data inconsistencies and errors are automatically detected and rectified . This saves time and prevents erroneous results. Thanks to a detailed visualization, it is possible to make a comparison between two models and their respective performances.

BigML
The main advantage of BigML is that it can be exported to any local server or can be instantly deployed as a real-time application.

Thanks to the “Partial Dependance Plots”, users benefit from a clear and intuitive interface allowing them to fully understand the predictive models .

Its easy-to-use web interface and REST API provide immediate access to users. In addition, the tool is available in a multi-user version, and in a Cloud or on-site version .

Google ML Kit
It is possible to create machine learning applications on mobile. Google ML Kit is a software development kit available on iOS and Android operating systems for smartphones and tablets.

This tool makes it easy to add machine learning functions to an application, without the need for machine learning or programming expertise.

Among the features offered are, for example, text recognition, face detection, landscape identification or barcode scanning. APIs also allow the use of TensorFlow Lite models on the smartphone application .

Little by little, details of the heirs of the Huawei P40 Pro and the rest of the family, in which we usually find part of the top exponents in Huawei photography each year, are being leaked. Let’s see what has been commented and shown about them.

What price will the Huawei P50 have?
The basic Huawei P (i.e. the intermediate model between the Lite and the Pro(s)) has been priced around €700 and €800 since the Huawei P20.

The one that changed with respect to its predecessor was the Huawei P40 Lite, whose starting price was 150 euros less than the Huawei P30 Lite, but seeing the trajectory it was something anecdotal (and with all the sense in the world to be able to compete more powerfully in the mid-range). There are still no rumors about prices, but from these precedents it is reasonable to think that this year’s prices would be along these lines.

What models will there be of Huawei P50?
Feature leaks are focusing on an alleged Huawei P50 Pro, but one would think that the brand would keep the traditional scheme, with which it is positioned in the mid, high and super-high ranges. Thus, we expect to see:

Huawei P50 Lite
Huawei P50
Huawei P50 Pro
Huawei P50 Pro+
The doubt above all is whether there will be Lite+ or if they will focus on boosting a single flagship in this first series of top of range. For now it seems that yes there will be P50 Por+ but there would be no P50 Lite, so it seems that the one that is more on the tightrope is the most basic one.

What is the design of Huawei P50 going to be like?
It seems that for those of Huawei curves remain the basis of its design and according to a leak of the front face of the Huawei P50 Pro this would have a screen with very curved edges and front camera with hole. We see that in this way the idea of the dual front camera that we saw in the P40 Pro would be abandoned, although above all it highlights that the hole would be more discreet.

Thus, at least talking about the Huawei P50 Pro we would see that the previous lines are maintained, taking the curve somewhat further in a supposed Pro+, with the goodbye to the island for the two cameras. In other leaks that also point to the hole in the screen for the front camera, differentiating between flat screen for the P50, curved (cascading) for the P50 Pro and more curved (Quad Curved, like this year’s) for the P50 Pro+. In other renders we also see that the rear camera module would maintain similar aesthetics to the previous one.

Will the Huawei P50 have 5G?
Most likely yes, seeing that this support is practically something already inherent in the high-end (with processors that mostly integrate the modem) and in view of the fact that the Kirin 9000 is hoisted as a very likely processor of the high-end P50. So it will remain to be known what bands it supports and also if it covers WiFi6.

Huawei P50: all the features we know about
The Huawei P40 Pro and Pro+ already drew a striking on the screen and from the leak that we discussed talking about the design it is deduced that at least the P50 Pro would have a metal outline ínfimo. The upper edge of the screen would house the front camera, leaking the following about the screens:

Huawei P50: 6.1-inch OLED withFullHD + resolution and 60 Hz refresh rate.
Huawei P50 Pro: 6.6-inch OLED display with QHD+ resolution and 90 Hz refresh rate.
If so, we would finally see a leap in resolution in the P (and not only in the Mate), something we did not see in the P40 or even in the Pro+ model. Of course, they would be more restrained in terms of refresh rate, remaining at a maximum of 90 Hz only in the Pro (perhaps also in a supposed Pro+).

The design leak also pointed to the dimensions of the phone. According to these data, the P50 Pro would be 159 millimeters high and 73 millimeters wide. The Huawei P40 Pro and Pro+ are 158.2 meters high and 72.6 millimeters wide, so everything points to maintaining the dimensions with respect to their predecessors.

In this render of the leak, the side power and volume buttons are integrated into the body of the mobile, probably made of aluminum. In addition, we could see a change in the sound output: from being located under the screen, it would change to an ordinary speaker in the upper area.

As for the processor, as we said when talking about 5G, the Kirin 9000, which we already saw in the Huawei Mate 40 Pro, is on the cards. Part of the bumps that the Chinese manufacturer saw on the road during 2020 by the US blockade was not being able to manufacture more of its own processors, but apparently they would have saved enough Kirin 9000 both for the P50 and already for the Mate 50 (foresighted manufacturer is worth two).

As for software, the P50s were most likely to have Android as a base with EMUI, the manufacturer’s own software layer, at least until the other day. The situation has not yet changed regarding the lock-in and for the time being they will not be able to have Google services, but according to Richar Yu said at the presentation of the Huawei Mate X2, HarmonyOS will arrive on Huawei flagships in April 2021. This would be getting ahead of its own roadmap, placing this in principle around October 2021, but we also do not know if it will stay only in the foldable or will come to the P50 as well.

In this sense, the last thing we saw before this ambiguous announcement is that EMUI 11.1 would be the software with which the Huawei P50 would start as standard and that it will be the last version of the layer, since Huawei will focus all its efforts on HarmonyOS. According to leaks, EMUI 11.1 would include the HarmonyOS kernel, a step that would facilitate the subsequent update to Huawei’s own firmware.

There would be versions starting from 6 GB of RAM (6 and 8 GB) and 128 GB of storage (128 and 256 GB). Of the battery there is talk of 4,200 or 4,300 mAh, with which we would not see a big jump with respect to the P40 and P40 Pro (4,200 mAh).

Yes it has been commented that we could see a 135 watt fast charge due to a leak. If so, it would surpass the 120W of the Xiaomi Mi 10 Ultra, although 65W or 50W seems more likely.

What are the Huawei P50 cameras going to be like?
There has been some less reliable leak showing an even larger rear camera module than we saw on the P40 Pro, keeping telephoto lenses. Other sources point out that they will premiere the liquid lenses that they talked about in GSM Arena for this year, but as we say it is not too reliable and would not provide too much data either, so we will have to wait to know more about this point, perhaps the most interesting of these phones.

What it does seem is that they will maintain the agreement with Leica that we see for the cameras of their high-end phones. So we will probably continue to see the seal of this company on the back of the P50.

What is the release date for the Huawei P50?

For now there are no rumors or mentions of changes to the roadmap or component production. There are also no announcements in the form of a teaser or poster from Huawei yet, so there is no reason to think that the usual schedule has any changes. The latest: there has been speculation around March 26-28.

That is why, if this is the case and if it keeps its usual pace, we expect to meet Huawei’s new P line next March. Last year the P40s were unveiled on March 26, so it fits the rumor.

The specialist in the sale of computer equipment LDLC opens a virtual reality studio dedicated to creating out-of-home entertainment experiences. The LDLC VR Studio’s first creation, “Catch the dragon”, sees two teams of 2 to 3 players compete in a 200 m² space in order to recover a dragon’s head as often as possible.

Another newcomer among virtual reality game creation studios! The LDLC group, specializing in the sale of computer equipment, announced Thursday January 16, 2020 the launch of LDLC VR Studio. He specializes in creating content for entertainment outside the home, and already markets his first game, called “Catch The Dragon”.

TWO TEAMS COMPETING

“Catch the dragon is both a cooperative and competitive game that takes place in a somewhat fantastic universe , ” explains Laurence Lacombe, director of information systems at LDLC. Two teams of 2 to 3 players compete for twenty minutes. The goal ? Collect as many times as possible the dragon head which is located in the middle of the arena.

Once equipped with a backpack computer, a VR headset and associated controllers, each player chooses their character class: a mage or an archer. The players are split into two camps: “Chaos” and “Order”. To make sure they dominate the game, they kill the characters on the other team. “The archer reproduces the real gesture of the archery , specifies Laurence Lacombe, and the mage draws energy using his stick” . After 10 minutes of play, the players swap sides. It is then possible to continue the experiment with the same character class or to change it.

A NEW GAME IN DEVELOPMENT

For now, “Catch the dragon” is only accessible at the MultiWorld area of ​​Bourgoin-Jallieu, in the Auvergne-Rhône-Alpes region at the price of 32 euros per person. LDLC VR Studio “is in discussions with other virtual reality rooms” , assures Laurence Lacombe. But this “free roaming” experience (in which players move physically) requires a space of 200 square meters, which limits the number of compatible establishments.

The virtual reality studio is already working on a new experience that could see the light of day in the second quarter of 2020. LDLC VR Studio seeks to “offer a completely different experience on a different surface” , specifies Laurence Lacombe, who will not say much more. . The studio, which is considering developing experiments in smaller spaces, could also make “static” virtual reality games in a standard space of 9 square meters.

UPCOMING RECRUITMENTS

The studio is currently made up of a team of six people based in the Lyon region. LDLC is however planning recruitments this year in order to give itself the means to achieve its ambitions. But LDLC VR Studio would not have quantified objectives for the moment, either for its sales or the number of productions to be produced. The studio does not close any doors and could even, why not, open its own room one day.

Huawei announces the arrival of Celia, its own voice assistant, in France in 2020. Already used on the Chinese market, the intelligent assistant should first make its debut on the brand’s smartphones. In a second step, artificial intelligence should replace Google Assistant and Amazon Alexa on all of the manufacturer’s devices, including connected speakers. 

As announced by Richard Yu, CEO of Huawei’s mobile division, in November 2018, Huawei will teach Xiaoyi new languages ​​to export it internationally . To prepare for Celia’s arrival in France and in the rest of the world, the firm has established an exclusive partnership with Qwant , Google’s rival search engine. Celia is therefore expected to rely on Qwant’s research results to answer your questions.

Collateral victim of the trade war between China and the United States, Huawei is no longer authorized to integrate Google services and applications on its smartphones. Private from the Google Assistant , we therefore expect the Chinese firm to integrate Celia on the future phones of its catalog under AOSP, the open source version of Android. Ultimately, Celia should also land on the Sound X, the premium speaker developed with Devialet .

Unsurprisingly, we imagine that the voice assistant will be added to Harmony OS, the alternative to Android developed by Huawei . As a reminder, the OS will be deployed on the group’s connected objects from 2020. Harmony OS is however not yet ready to supplant Android on a smartphone , the manufacturer recently admitted. We will tell you more as soon as possible.

Head to CES in Las Vegas for Ido-data, a start-up from Lyon that publishes Dial, an individual alert and localization solution at sea for those practicing nautical activities. With its futuristic design imagined by the famous Philippe Stark, the bracelet should seduce lovers of the waves and the beautiful.

Winner of several innovation awards including Graines de Boss, the 5th edition of the French IoT or the Victoires de l’ Innovation , the start-up created in August 2016 is already on the rise. ” At the start, the two founders Antonin Carlesso and Yannick Toquet had in mind to create a bracelet to prevent children from getting lost on the beaches , says Clara Fresnel, marketing and communication manager at Ido-data, then exchanging with rescuers at sea – Yannick being a spearfisher – they quickly wanted to expand to a wider audience . ”

Tested all summer by a thousand rescuers at sea from the SNSM (Société Nationale de Sauvetage en Mer) spread over 260 rescue stations in France, the bracelet contains a tag connected to a multi-operator SIM card, capable of choosing the best network depending on its location, up to more than 20 km from the coast. It is enough for the practitioner of paddle, kitesurf, windsurfing or even small coastal yachting, to choose a referent person, and to this one to install the application on his mobile phone. Once the alert has been triggered, a notification is sent to the application and the referent can in one click share the information with the nearest aid station. ” Dial allows you to be found in less than 30 minutes, the system divides the average rescue time by three”, notes Clara Fresnel.

A FIRST STOCK OF 8,500 DEVICES

Today, the solution is marketed in 34 countries and the French overseas departments via the online lifeguard shop, the specialized e-commerce site, Surfshop or Amazon . A first stock of 8,500 products was manufactured in China by the Orléans-based industrialist Altyor, at an ex-factory cost of € 60. ” Secondly, we are thinking of repatriating production to Europe”says Clara Fresnel. The 149 € pack contains the bracelet, the tag and an induction charger and one year of subscription to the device. Each additional year of subscription is offered at a price of 29 €. A great lover of the sea, the designer Philippe Stark wished to join this project on a voluntary basis to begin with, then part of the fees will be donated to the SNSM, once the start-up has found its cruising speed.

FUNDRAISING OBJECTIVE

The team will go to the CES from January 7 to 10, 2020 with logistical and financial support from La Poste . A capital opening of up to 300,000 euros is planned to strengthen the sales team. ” Our aim at CES is to find potential investors and to build relationships with distributors like Décathlon or Le Vieux Campeur, in France and in Europe as a priority “, book Clara Fresnel. For 2020, the start-up aims to produce 20,000 devices in France, Spain, Italy and Germany . It will also offer an offer to professionals such as boat and windsurf rental companies for example.

The French Red Cross is modernizing its nursing training institutes. She will deploy ten digital simulation modules to support her students, including two in virtual reality. She is also studying the potential of virtual reality for therapeutic support and public awareness.

The training of healthcare professionals is undergoing a real transformation thanks to digital technology. The French Red Cross has understood this, and will deploy ten digital simulation modules – including two in virtual reality – in several of its training centers.

It must be said that with 36 establishments in France, including 34 nursing training institutes (IFSI) and 33 nursing assistant training institutes (IFAS), it is a major player in this sector. Each year, it trains nearly 20,000 students in initial training and 145,000 professionals in continuing education. This deployment was announced by Jérôme Leleu, CEO of the Bordeaux company SimForHealth, on December 10, 2019 during the New Tech Day event organized by L’Usine Digitale.

SIMULATORS DESIGNED WITH THE NOUVELLE-AQUITAINE REGION

This collection of training modules was indeed developed by SimForHealth with the support of the Nouvelle-Aquitaine region. It has already been deployed in the 27 IFSIs in the region, and the genesis of the project is one of the factors that attracted the Red Cross. The collection was co-designed with around forty trainers from the IFSIs to ensure that they meet their needs, but also with the regional health agency to validate the educational soundness of the project.

“We approached the question by skills and not just technical gestures, based on the repository of 10 nursing skills at the national level , explains Jérôme Leleu, CEO of SimForHealth. We also chose to work on a complete collection of training modules, which meet all the needs of a course, rather than limiting ourselves to fragmented experiences. ” 10,000 students are likely to use this platform to u within the institutions of the French Red Cross.

AN INTEREST IN THERAPEUTIC SUPPORT AND AWARENESS

This initiative is part of a more global approach by the Red Cross around virtual reality. In particular, it has created a dedicated VR space within 21, its social innovation accelerator . “Immersive technologies have great potential in three main areas of action for the French Red Cross: training, therapeutic support and awareness-raising, explains Marie Alméras, assistant to the director of strategy and innovation. By creating a virtual reality studio in our common home, we are offering all of our players to use these technologies and derive the greatest benefit from them in the service of their mission. “

To launch its car-sharing vehicle service called Mocean Carshare, Hyundai has moved closer to Frenchman Vulog and its AiMA platform. It is designed to handle different types of vehicles, services and users. A boon for Hyundai, which is already planning to roll out other mobility services in the future.

Hyundai follows suit with many of its competitors. The Korean manufacturer announced Thursday, November 14, 2019 establish an entity dedicated to mobility services that is called MoceanLab. This new entity will launch a first car-sharing service in Los Angeles, USA .

UP TO 300 VEHICLES IN CAR-SHARING

Twenty hybrid Hyundai Ioniq Plug-in vehicles will be available to users by the end of the year on the Mocean Carshare platform. Then Hyundai wants to gradually deploy a fleet of 300 vehicles. The Korean manufacturer has received a permit from the Los Angeles Transportation Department to operate this service.

Hyundai says it chose Los Angeles because of its size, high population density, and significant congestion problems. In order to solve these mobility problems, the city is turning to innovations in this field, whether in terms of micro-mobility or multimodal transport.

OF SELF-SHARING AT THE SERVICE OF ROBOT TAXI

MoceanLab has come closer to the French Vulog and its AiMA car-sharing platform. This can be used to deploy a car-sharing service, but also to offer taxi robots or scooters. Right from the start, Vulog designed its platform to handle different types of vehicles, services and users. A boon for MoceanLab who wants to extend its offering far beyond a car-sharing vehicle service in the run-up to the Los Angeles Olympic Games in 2028.

In the long term, Hyundai says it wants to offer mobility solutions that can connect people to public transport hubs using robot taxi systems, autonomous shuttles, multimodal transport and personal mobility. “Hyundai Motor Group will develop an innovative mobility platform and offer various mobility services in Los Angeles,” Kyounglim Yun, executive vice president of Hyundai, said in a statement.

HYUNDAI IS WORKING HARD

The Korean manufacturer has recently affirmed its ambitions in the field of autonomous vehicles and mobility. Hyundai will invest $ 35 billion by 2025 in the development of autonomous, connected and electric vehicles and technologies around car-sharing.

In September, the Korean announced a joint venture with Aptiv to develop a platform for autonomous vehicles. And a month later, Hyundai launched an on-demand taxi robot driver named BotRide in partnership with Via and Pony.ai.

Lenovo took advantage of its Tech World Conference in Beijing to unveil a prototype of augmented reality glasses. The accessory, whose production is not guaranteed, has been designed to reinforce the visual isolation of people working on the move. It has the distinction of working by connecting to a computer.

This is an umpteenth name on the list of manufacturers to take an interest in augmented reality glasses. Lenovo took advantage of its Tech World Conference, held in Beijing on November 14, 2019, to announce a prototype called Lenovo AR Concept Glasses. To stand out from the competition, Lenovo has chosen to focus on a niche audience: professionals on the move.

The glasses must be connected to a PC by a cable, the main use case put forward is the possibility of seeing multiple screens in a small space. Another interest: they prevent any indiscretion from those around the wearer. Just out of the lab, the accessory could help reinforce visual isolation in public spaces… and improve productivity.

VIEW MULTIPLE SCREENS

If the technical aspects of the possible future product have not yet been revealed, it would ship three front-end sensors according to the official photos of Lenovo. It is closer to the glasses presented by his competitor Nreal at the last CES, held in January in Las Vegas ( United States ).

The company has not yet announced the production of its augmented reality glasses. She said, however, that the virtual monitor is only “one of the cases of use to study” and that “new features will happen very soon”. Such a range of products would complement the first escapades of the Chinese manufacturer in augmented reality. The company recently announced a new version of its consumer headset, the Lenovo Mirage AR. In B2B, the ThinkReality A6 was unveiled a few months earlier – a pseudo-competitor HoloLens.

Augmented reality glasses are part of the major projects of manufacturers for the next 5 years. Some rumors report that Apple would have made a strategic focus for the coming years, going so far as to think that they will make an ideal successor to the iPhone by 2023. Google, who arrived too early on the niche with his Google Glass in the early 2010s, is obviously part of it. Microsoft and Facebook are also working on these products.

It is these systems, qualified by Yann LeCun, Chief AI Scientist at Facebook , ” the most interesting idea of ​​the last 10 years in the field of Machine Learning “. which today allow an ever closer resemblance between the false and the true. Introduced in 2014 by the American researcher Ian Goodfellow, they have the particularity of being able to create data (for example images in the case of deepfakes). And this without their learning being supervised.

How? “The GANs are made up of two competing neural networks: the generator, which aims to create images as realistic as possible, and the discriminator, responsible for recognizing whether or not the images produced by the generator are fake” , explains Elisa Fromont, a researcher specializing in machine learning at the Institute for Research in Computer Science and Random Systems (IRISA).

In fact, we start by feeding the data generator including “noise” (random images), which it will try to extract itself, by encoding, some rules. These will allow him, by decoding them, to produce images. “Initially, the generator will not produce anything very interesting since its first images will look like the noise it was given in input,” says the researcher.

LEARN TO REPRODUCE A STYLE

To make it progress, each new image created will be submitted, in the middle of a stream of other images, to the examination of the discriminator – a conventional convolutional neural network – which, for its part, will compare it to the catalog. real images put at its disposal and give a probability of its authenticity between 0 (a fake) and 1 (a real image). This estimate is returned to the generator who, with the new data, produces a new image that is always closer to the catalog images that serve as a reference.

The goal is not to literally copy the real images but to create new “in the manner of” the latter. The two entities pursue this way, each participating in the improvement of the other. Because if the generator learns to make images more and more realistic, the discriminator, for its part, learns to distinguish better and better the real images of false. “It is quite difficult to train such networks because the discriminator must be good enough so that the images of the generator can look like real images,” raises Elisa Fromont. The images are considered realistic when the generator manages to deceive the discriminator. The resulting false is then supposed to be almost undetectable.

Ford has found its next test site for its autonomous vehicles: the city of Austin in Texas, USA. The automaker said Wednesday, September 25, 2019 that autonomous vehicles will initially circulate in manual mode to properly map the streets of the city and take its pulse.

After Miami and Washington DC , Austin (Texas, USA ) is the third market on which Ford will test its autonomous vehicles. The announcement of these tests, conducted in collaboration with Argo AI, was made by the manufacturer Wednesday, September 25, 2019. Ford is working closely with this start-up in which he announced wanting to invest $ 1 billion in 2017 . Argo AI uses machine learning to develop a reconnaissance system so that the autonomous vehicle detects pedestrians and other vehicles around them.

AUTONOMOUS VEHICLES IN MANUAL MODE

At first, autonomous vehicles will circulate in manual mode east of the city and in the city center. Sensors mounted on the vehicle allow for 3D and HD mapping of streets, buildings and static and permanent objects. It also gives them a clear understanding of the environment in which autonomous vehicles will evolve and to gain insight into the situations they will face. In Austin there are many pedestrians, cyclists and electric scooters that can complicate the situations encountered by the autonomous vehicle. In a second time, the tests will be extended to other parts of the city.

However, Ford does not specify the number of autonomous vehicles that it plans to deploy in the city nor when it hopes to achieve real tests in autonomous mode. Beyond the development of the technology itself, the automaker also seeks to understand how it can best deploy its autonomous vehicles in the city. Ford says that before launching such a service, he needs to understand the benefits that local residents and entrepreneurs will derive from it.

Austin is an interesting field of experimentation as the city grows and the number of vehicles used in this city could double by 2040 while the capacity of roads would only increase by 15%, according to Ford. The American company wants to find how a fleet of autonomous vehicles can at best fit into the city, whether to offer a service of folding people to transport or a home delivery service.

On September 24, 2019, Google made available a database containing 3000 deepfakes. The goal is to help researchers who design tools to fight against real-life video fakes that can be very dangerous during an election period.

The spectrum of deepfakes haunts the upcoming US presidential elections in November 2020. Indeed, these true-to-life video fakes are very difficult to detect and represent a real danger to the integrity of information during an election period. Without great technical knowledge, it is possible from a limited sample to imitate the face and voice of anyone. Google understood this problem and decided to act.

PROVIDE CONTENT FOR ACADEMIC RESEARCH

On September 24, 2019, the digital giant has made available a set of data containing 3000 deepfakes for researchers specialized in their detection. Thus, they will be able to train their tools more easily.

To create this set, Google has filmed actors with their consents and paid for them. He also promises to regularly feed this database to keep up with rapidly changing technologies in this area. The latter has been integrated in “FaceForensics benchmark”, a program of detection of manipulated videos designed by the Technical University of Munich and Federico II University of Naples.

EXAMPLES FLOURISH ON THE CANVAS

Donald Trump accusing Barack Obama of rape, the former president of the United States treating his successor “asshole” or Mark Zuckerberg taking a funny position on privacy … these examples flourish on the web. And deepfakes can go even further, to pornography for example. In the autumn of 2017, for example, a video features British actress Daisy Ridley on the Reddit community site. It is easy to understand the harm that may arise from such a situation.

A SIMILAR INITIATIVE LAUNCHED BY FACEBOOK

A similar initiative was launched on September 6, 2019, by Facebook, Microsoft and several prestigious universities. The social network commissioned the creation of a set of data that was made available to researchers and then launched a deepfake detection contest with the key $ 10 million prize. In the sight of these actors: the next US elections in November 2020.

ANTI-DEEPFAKE LEGISLATION IN THE UNITED STATES

According to the American website The Verge, Texas would have a law “anti-deepfakes” especially for the next elections. The latter came into effect on September 1, 2019.

For its part, the Hexagon has not yet had specific legislation for deepfakes. A first step was still taken with the law of 22 December 2018 on the fight against the manipulation of information. In concrete terms, this text created a new route of summary proceedings aimed at stopping the dissemination of false information in the three months preceding a national election. Once seized, the judge must assess, within 48 hours, whether these data are disseminated “artificially or automated” and “massive”.

Tencent, like the other giants of the web Alibaba and Baidu, prefers, according to the experts, to invest in the application of virtual reality to e-commerce, which allows a customer to simulate fitting a product before buying it. line.

These groups are also eyeing entertainment content in virtual reality – to give the impression of attending a concert just by wearing glasses.

For their part, local authorities are promoting new uses for these burgeoning technologies: a number of cities in China are actively supporting incubators in this field and integrating virtual reality into their research and education programs.

Seekers VR, which has a franchise network of 200 arcades in more than 70 cities in China, is working with the city of Wenzhou (east) to create an institution where students will be trained in virtual reality, while integrating this technology into their courses.

“This market is still immature, there is not yet a dominant player.We will offer more and more opportunities,” promises the CEO of the company, Belle Chen.

In the coming years, virtual reality could benefit fully from the expected deployment of 5G, the ultra-fast fifth generation of mobile internet that has revolutionized communications – but also sectors such as education, abounds Chen Wei, manager. of the Shanghai arcade hall.

For him, virtual reality is the most effective and least expensive way to train. “We are still explaining what it is, but one day, it will explode.”