Connect with us

Big Data

Microsoft and corporate activism

Avatar

Published

on

BOSTON — Artificial intelligence needs to eat, breathe and sleep data to be effective. By that measure, the company…

— or country — with the most data should emerge victorious. But Brad Smith, president and chief legal officer at Microsoft, said the idea that data accumulators will rule the world and the rest of us will be powerless is a pessimistic take on what’s to come.

He prefers optimism. Indeed, Smith said there are ways to keep, say, China from becoming a global data dictator. His recommendation: Create “a global standard for ethical principles and for the protection of things like privacy so that the price for global admission is adherence to a global standard.”

Such a standard could create a classic data-silo problem for countries that don’t comply. For example, China may have the world’s largest population, but if it can’t access Europe’s data or the United States’ data, it will struggle to uncover patterns on a global, rather than national, scale.

That was one of the points Smith made at HUBweek, an innovation festival in Boston. In a fireside chat with Adi Ignatius, editor in chief of Harvard Business Review, Smith made it clear that Microsoft is participating in a new kind of corporate activism that targets broad societal issues — and it’s using this newly public corporate conscience as an advocacy and marketing tool.

For example, in December, Microsoft publicly supported a giving women and men the right to take a sexual harassment claim to court rather than keep the complaint in arbitration. When North Carolina a that restricted LGBT rights, Microsoft lent its voice to the opposition. More recently, the company launched the Defending Democracy Program, which is aimed at protecting campaigns from hacking, increasing political advertising transparency and defending against disinformation.

The political thread in the examples Smith provided is hardly accidental: Customers are increasingly turning to companies to take on issues they care about because of the dysfunction in Washington, D.C., Smith argued.

“We are living in a time when there are historically low levels of trust in government,” he said. “And so, whereas in the past, people would say, ‘I care about this. I’m going to go to government.’ They are less inclined to do that.”

But jumping into corporate activism and promoting a company’s moral compass can also create wrinkles. Smith said, given the gridlock in Washington, customers are beginning to ask Microsoft to help regulate the very technology it’s developing. One HUBweek attendee raised a question that has become common at tech events: Can government officials craft policy that effectively keeps up with the rapid pace of change in the tech industry? 

Smith shot down the idea, saying that any technology company’s involvement in regulation is inappropriate and restricts progress. “I don’t think it’s viable to ask tech to slow down,” he said. “It is not only appropriate, but it is right to ask government to move faster. And it’s incumbent on those of us in the technology sector to do what we can to share information so that governments can move faster.”

Plus, he said, “it’s important to remember that, in this country and in many others, people elect those who make the laws. People do not elect companies.”

‘Moneyball’ for movies

Legendary Entertainment is using analytics to develop films. Matt Marolda, chief analytics officer for the media company, referred to this as “Moneyball for movies,” referencing the story of how data helped transform a losing baseball team.

The use of data to develop movie products isn’t , but in the past, much of the data was collected through analog methods. These days, the data that Marolda’s applied analytics team uses for analysis comes from a variety of sources — from one-on-one conversations with viewers to digital data such as search queries and Twitter conversations.

The tried-and-true method of testing a film before a live audience is also used — but with a twist. While Marolda’s team tries “to be as unintrusive as possible” during the testing process, it uses iPads to capture facial expressions and wristbands to collect heart rate and other “various signals,” he said. That kind of data is used to determine what’s working and what’s not.

“We try and identify those moments where people are confused, people are rolling their eyes, where they’re really engaged,” Marolda said.

The data can also help pin down much bigger problems such as whether the chemistry between the two leads is believable. Marolda said filmmakers often have a gut answer to human chemistry questions like the latter, but the data can provide solid evidence.

“With that kind of evidence, entire storylines are removed from movies, reshoots might happen to redirect the plot,” he said. “Those indicators are a great way to understand what the opportunities are for improvement.”

Coinsmart. Beste Bitcoin-Börse in Europa
Source: https://searchcio.techtarget.com/news/252450811/Microsoft-and-corporate-activism

Big Data

Colonial Pipeline hackers stole data on Thursday – Bloomberg News

Avatar

Published

on

(Reuters) – The hackers who caused Colonial Pipeline to shut down on Friday began their cyberattack against the top U.S. fuel pipeline operator a day earlier and stole a large amount of data, Bloomberg News reported citing people familiar with the matter.

The attackers are part of a cybercrime group called DarkSide and took nearly 100 gigabytes of data out of Colonial’s network in just two hours on Thursday, Bloomberg reported https://www.bloomberg.com/news/articles/2021-05-09/colonial-hackers-stole-data-thursday-ahead-of-pipeline-shutdown?sref=SCAzRb9t late Saturday, citing two people involved in the company’s investigation.

Colonial did not immediately reply to an email from Reuters seeking comment outside usual U.S. business hours.

Colonial Pipeline shut its entire network, the source of nearly half of the U.S. East Coast’s fuel supply, after a cyber attack that involved ransomware.

(Reporting by Aakriti Bhalla in Bengaluru; Editing by Himani Sarkar)

Image Credit: Reuters

Coinsmart. Beste Bitcoin-Börse in Europa
Source: https://datafloq.com/read/colonial-pipeline-hackers-stole-data-thursday-bloomberg-news/14521

Continue Reading

Big Data

Musk boosts his brand, and NBCUniversal’s, on ‘Saturday Night Live’

Avatar

Published

on

(Corrects word in quote in paragraph six, and spelling of mother’s name in paragraph seven)

By Joseph White

DETROIT (Reuters) -Billionaire Elon Musk dropped a surprise early in his hotly anticipated turn as host of “Saturday Night Live,” saying in his monologue that he “is the first person with Asperger’s” to host the show, before clowning through skits for the first global livestream of the NBCUniversal comedy show.

Musk, one of the world’s richest individuals, opened his monologue by telling an audience in more than 100 countries he is “the first person with Asperger’s to host SNL. At least the first to admit it.” The billionaire made light of his tendency to speak in a monotone, adding “I’m pretty good at running human in emulation mode.”

Asperger syndrome is a condition on the autism spectrum that is associated with difficulty in social interaction, and sometimes is referred to as high functioning autism.

Comedian and Saturday Night Live alumni Dan Aykroyd has spoken in interviews about being diagnosed with a mild form of Asperger’s. He hosted Saturday Night Live in 2003.

Many had wondered how Musk would handle himself during a live comedy show. The answer was that he, and the show’s writers, sought to soften the rough edges of Musk’s public persona. Throughout the show, Musk gently poked fun at himself, including his penchant for provocative tweets and the time he smoked a joint on a podcast.

“To anyone I’ve offended I just want to say, I reinvented electric cars and I’m sending people to Mars on a rocket ship. Did you think I would be a chill, normal dude?”

Musk’s mother, Maye, joined him on stage and the two made a joke about Dogecoin, the cryptocurrency Musk has touted. Cryptocurrency jokes popped up throughout the show. In one sketch, Musk was cast as a bow-tie wearing cryptocurrency expert on the show’s Weekend Update segment. Dogecoin and other digital currencies had surged in price ahead of Musk’s SNL appearance.

Dogecoin fell during the show and was last down about 12% from late Friday at around $0.53. It hit a record high Thursday above $0.73.

Musk was most convincing playing a version of himself as head of SpaceX dealing with an emergency on a Martian colony. The crisis had a happy ending, until it didn’t.

In the end, Musk will keep his day jobs. Still, the “Saturday Night Live” appearance offered plenty of synergies with his real gigs as “technoking” and Chief Executive of Tesla Inc, head of rocket launch company SpaceX and even chief of the Boring Company, a tunnel construction venture.

Musk got days of attention across all forms of media ahead of the show, and shared the spotlight with a prototype of Tesla’s futuristic Cybertruck that Tesla brought to Manhattan on Friday. Video of the hulking, angular pickup prowling Manhattan streets blew up on social media.

During the show, a Tesla supercharger made an incongruous cameo appearance in a skit set in an old West saloon. Musk played a gunslinger who had developed an electric horse, and advocated tunneling through the earth to escape a shootout.

Musk often boasts that Tesla doesn’t spend billions on advertising the way established automakers do. He doesn’t have to so long as he has access to platforms like Twitter or “Saturday Night Live.”

Musk’s appearance also boosted NBCUniversal. The media company used Musk’s global celebrity – and the controversy surrounding his appearance on a stage normally reserved for film stars or professional comedians – to get attention for launching the “Saturday Night Live” franchise beyond the confines of broadcast television. The company said Saturday’s show was streamed live via Alphabet Inc’s YouTube to more than 100 countries.

(Reporting By Joe White; Editing by Simon Cameron-Moore)

Image Credit: Reuters

Coinsmart. Beste Bitcoin-Börse in Europa
Source: https://datafloq.com/read/musk-boosts-brand-nbcuniversals-saturday-night-live/14520

Continue Reading

Artificial Intelligence

The future of e-commerce: Trends, tips, traps to avoid

Avatar

Published

on

Amazon is approaching its 30th anniversary, set to mark the milestone in 2024. The World Wide Web hits 35 the same year. E-commerce, the buying and selling of goods and services over the internet, has grown up — and it has gotten big. Worldwide e-commerce sales for the retail sector alone exceeded $4 trillion in 2020, according to eMarketer. The research firm expects the figure to hit $5 trillion in 2022. Global B2B e-commerce sales, meanwhile, hovered around $6.6 trillion in 2020, according to research firm Frost & Sullivan.

As the value of e-commerce has risen, so has the complexity of online transactions. E-commerce today means more than simply processing electronic payments and enabling internet sales. It’s also more than knowing your customer, crucial as that is. E-commerce sales in 2021 depend upon the robust performance of just about every aspect of modern enterprises, from operations and supply chain to delivery services and customer loyalty programs.

Organizations must harness all the power of integrated back-office systems in tandem with intelligent customer insight systems to deliver personalized, seamless digital transactions that — in the lingo of the age — delight the customer. Personalized, seamless transactions must happen whether the customer is an individual consumer buying his or her first product, or a global business ordering for the 100th time under a multiyear procurement contract. Buyers demand as much — whether they’re ordering from their computer or their smartphone, via Alexa or through another connected machine.

“E-commerce transactions are becoming ubiquitous, and expectations are going up. People have expectations that it’s always going to be as easy as using Uber to get a ride,” said Mike Welsh, chief creative officer at Mobiquity, a digital consulting agency.

Amid rising customer expectations, however, many organizations are falling short on their e-commerce operations. A Gartner report on the COVID-19 pandemic’s impact on digital commerce predicted that “through 2020, 50% of large organizations will have failed to unify engagement channels, resulting in a disjointed and siloed customer experience that lacks context.”

The bar is high, said Lisa Woodley, vice president of customer experience at NTT Data Services. “E-commerce [covers] every stage, from acquisition to loyalty and advocacy. It’s your customers telling their friends, ‘I had a great experience; go do business with this company.’”

In this look at the future of e-commerce, we examine the evolution of buying and selling over the internet — from the early corporate websites that functioned as online brochures to today’s powerful, concierge selling sites that can be accessed through multiple channels. We offer expert analysis of the impact of COVID-19 on digital transactions, delve into the challenges enterprises face in meeting customer expectations in 2021, and provide detailed advice on overcoming those challenges.

From ‘product-centric’ to ‘solution-centric,’ e-commerce evolves

A combination of factors is driving the evolution of e-commerce. At the core is the internet.

Companies once mostly competed on the so-called four P’s of marketing: place, price, product and promotion. But the web’s search function and the internet’s reach neutralize one or more of these differentiating factors. Shopping online, a customer can easily get the same or similar product at the same or lower price with comparable shipping times and costs.

As a result, other factors are emerging as key differentiators, with personalization being the catch-all term for the new elements that drive buying habits in the digital realm.

“The concept of e-commerce is shifting from online sale transactions [and a] static webpage to a personalized and interactive experience,” said Eleftheria Kouri, a research analyst with the tech market advisory firm ABI Research.

“Customers have access to a wider range of capabilities when visiting an online store, including product virtual try-on and gaming and interactive storytelling concepts that increase engagement and educate the consumer about products [and] brands.”

Penny Gillespie, vice president at Gartner and a fellow in its customer experience/digital commerce team, said that in the e-commerce marketplace of 2021, companies must figure out how to deliver the product and the solution to a customer’s problem. To do that, they must understand the online customer’s intent.

For example, a retailer serving a customer searching for a black dress should be capable of using digital tools, as well as general and personal data, to understand that the shopper doesn’t simply need a dress but rather needs an outfit for an event. In fact, the color of the dress may in this case be irrelevant — with black dress being nearly synonymous with cocktail dress.

“Understanding intent is part of personalizing an experience,” Gillespie explained. A retailer that understands this concept can ensure the products in the search results actually match that customer’s needs, guaranteeing the sale of a dress and other relevant items (e.g., accessories) — and ensuring repeat business.

Customer intent is relevant in B2B transactions as well. Here, it could mean understanding a customer’s unique procurement process by, for example, automatically displaying any special prices specified by an existing procurement contract, facilitating any approval requirements, and anticipating needs based on past ordering histories.

“It’s a work in progress, with some sellers being much better at it than others,” Gillespie said.

In both the B2C and B2B spaces, online selling has gone from being reactive to being proactive and participatory, said Gillespie: “It’s a move from being product-centric to solution-centric.” She used the sale of an exercise bike online as an example.

“It’s not just selling an exercise bike online, but rather delivering it to the buyer’s house, setting it up and then helping them maximize its value through use,” she said. “The bike is a product; when it’s in my house and working, it is the solution.”

COVID-19 pushes companies and customers into the digital realm

The evolution of e-commerce from static webpages to interactive customer “solution” sites was enabled by sophisticated technology, but it took a global health crisis to make the future of e-commerce the new normal. The shift to online-everything in 2020 due to pandemic-induced social restrictions and quarantine orders pushed physical transactions into the digital realm.

According to findings from consulting firm McKinsey & Company, e-commerce as a percentage of overall retail sales in the U.S. grew 3.3 times more in 2020 than the average annual rate in five years before COVID-19. E-commerce sales as a share of overall retail sales grew 4.6% in 2020 vs. an average of 1.4% growth in previous years.

“Consumers are demanding more digital access than ever before,” said Nicole West, vice president of digital strategy and product at Chipotle Mexican Grill.

In November, the restaurant chain opened its first “digital-only restaurant,” the Chipotle Digital Kitchen, in Highland Falls, N.Y. The location offers pickup and delivery only, a prototype the company said will allow Chipotle to enter more urban areas that don’t support its full-size restaurant concept. The new restaurant requires customers to order in advance via Chipotle.com, its app or through third-party delivery partner platforms.

Providing an exceptional digital experience has become a priority for the 28-year-old chain, West said. She added that Chipotle is “relentless when it comes to UX and making it fast, easy and convenient” to place digital orders.

She cited the company’s 2020 rollout of Unlimited Customization. A feature in the Chipotle app and on the company’s website, it allows customers to customize orders, just as they do when ordering in person at a restaurant. Earlier in 2020, the company launched ordering on Facebook Messenger and a Group Ordering feature, which allows multiple people to participate in the ordering process simultaneously on the Chipotle app and Chipotle.com. And it’s now testing Chipotle Carside at 29 restaurants in California, an in-app feature that lets customers have their Chipotle orders delivered to their parked cars.

Chipotle’s digitalization efforts have shown real-world business value. Digital sales for Chipotle have grown 177% year over year, West said, and they accounted for 49% of sales in the last quarter of 2020. More than 19 million people joined the company’s customer rewards program via digital sales, West added, noting that the company’s digital pickup orders are currently its most lucrative transaction type.

E-commerce trends

The appetite for digital access is unlikely to abate. The McKinsey report cited above noted that approximately three-quarters of people who used digital channels for the first time during the pandemic said they plan to continue using them when normalcy returns.

There is no denying that the COVID-19 crisis and the at-home new norm have reshaped consumer behavior and boosted e-commerce/online shopping, which is expected to continue growing after the end of the pandemic, ABI Research’s Kouri said.

The technology powering these e-commerce trends also continues to evolve rapidly, Kouri noted, citing technological advancements in smartphones — such as high-resolution cameras and displays — enhanced connectivity, mobile-friendly websites and the rise of social media shopping.

Amazon, of course, has continued to make online shopping easier with innovations such as its Add to Cart and Buy Now buttons. The Home Depot and Lowe’s are often lauded for their use of instructional videos that give customers confidence to make purchases, as well as for apps that help customers navigate their stores. And the use of various technologies to let customers see how their items will look on them or in their homes before they buy is becoming standard practice.

From the customer’s perspective, Gartner’s Gillespie noted, the benchmark for all digital transactions is “the last great experience they had.” Keeping up with that moving target will require a panoply of technology and continuous technology innovation.

E-commerce technologies

Although the internet was the enabling technology for e-commerce, it is far from the only technology needed to deliver the experience that customers expect now and moving forward. Some of these broad technology capabilities include the following:

  • Customer-facing capabilities. Sites must be easy to navigate and user-friendly as well as quick and responsive. Sites should have the features that matter most to the target audience and be able to interact with other sites — social media sites for young consumers, for example, or company procurement systems for corporate customers.
  • Data-related technologies. Organizations must be capable of collecting and using their own data as well as data from outside sources. This allows the organization to anticipate a customer’s needs even when it has little or no data on that specific customer; the company can use its other data sources to compile an understanding of what that one customer needs based on its interactions with similar customers.
  • Automation technologies such as RPA. Robotic process automation can speed and streamline processes that service the customer by minimizing errors in data collection, enabling self-service by providing access to back-end systems.
  • Customer journey orchestration engine software. This class of tools help organizations analyze real-time data of individual customers to predict future interactions with that customer, using predictive models, decision trees, matrix rules and machine learning.
  • Augmented reality. AR lets customers bring products into real lifelike situations and virtually try on or fit items before purchasing. “The introduction of digital tools, such as augmented reality, in e-commerce platforms or apps not only assists brands to differentiate from the competition but transfers static websites/2D images to interactive and personalized experiences,” Kouri said.
  • Artificial intelligence. Organizations can use AI to offer personalized online experiences. A cosmetics brand, for instance, could use AI algorithms to provide skin analysis and recommend suitable products.
  • Back-end systems. Companies need modern infrastructure and current IT architecture that can support all these other capabilities. Typically, this means moving from legacy systems to cloud computing and SaaS applications to quickly enable scale and speed when needed; leveraging microservices to increase agility and flexibility; and breaking down silos through integration and the use of APIs. “There’s actually a lot more on the back end needed to reach our goal of making the front-end experience as seamless as possible,” Woodley said.

Specific tools, such as geofencing platforms that provide location-based services to help organizations and customers pinpoint their locations, and payment systems also have an important role in an e-commerce strategy, as do the technologies and processes companies use to optimize their warehouse and supply chain management.

“Building competitive e-commerce experiences requires the synergy of numerous technologies and tools, from AR to AI and secure payment systems,” summed up Kouri.

E-commerce challenges

Bringing all these parts together to work consistently and flawlessly is, not surprisingly, a significant challenge.

“Personalization is not a one-size-fits-all approach. You really need to consider your business model, value proposition and customers before you create your strategy. Once those pieces are solidified, you can then begin to seek out the right tools and technologies needed to be successful,” said Britt Mills, senior director of customer experience at Mobiquity.

Organizations also need the data experts, technologists, marketing team, logistics workers, supply chain personnel and other executives and supporting staff who can competently contribute to that vision.

“Stores can rush to market with a new technology to enable customer safety and convenience, but they shouldn’t do so at the customer’s expense,” Mills said. Training employees to use the new technology is essential. “If your associates can’t support this new expected experience, your customers won’t be satisfied. It doesn’t matter how great the technology is.”

In addition, companies must have a strategy for dealing with emerging data use laws that put more control over personal information into the hands of individuals. And they must be able to mitigate against escalating cybersecurity risks.

These capabilities and safeguards are hard to achieve. Experts have acknowledged that the difficulty of developing and implementing know-your-customer processes — from collecting the necessary data to analyzing it to turning it into action items — has been oversimplified and glossed over in many conversations.

It’s not shocking, then, to learn that most organizations are struggling to develop the capabilities required to deliver seamless, personalized service, especially as the number of engagement and delivery channels have increased.

Research from Verint, a provider of customer engagement management products, found that 82% of the nearly 2,300 business leaders it surveyed said the challenges of managing customer engagement will increase in 2021, but only 50% said they’re well prepared to support customer engagement priorities moving forward. The vast majority of those surveyed pointed to nearly every aspect of customer engagement as challenging for their organizations, indicating the following problems:

  • understanding and acting on rapidly changing customer behaviors (94% cited);
  • managing the growth in volume of customer interactions (88%);
  • achieving a unified view of customer engagement and overcoming data silos (79%);
  • using customer feedback to improve experiences (78%); and
  • building enduring customer relationships (77%).

Customer journey mapping

How do traditional organizations become as competitive in the digital sphere as they were in the brick-and-mortar heyday? It starts with mastering customer journey mapping, according to Peter Charness, vice president of retail strategy for UST, a digital technology and transformation IT services and solutions firm.

“[Organizations] need to ensure the digital journey is well aligned to a shopper’s needs, using a high degree of personalization and creating relevant interactions and conversations,” Charness said. He laid out six areas where organizations need to benchmark their capability:

  • Interest generation. Determine how successful your organization is at getting potential customers to its website or store.
  • Research and decision influence. Examine whether it’s easy for the user to find products of interest and “gather the information and confidence they need to move that product into a shopping cart,” Charness said.
  • Decision confidence. From browsing to buying, companies should make it easy for the shopper to say yes to a purchase. “Organizations should have this part of the conversation with their shoppers and build their confidence to press ‘buy,’”Charness noted.
  • Delivery/collection. “Speed of delivery (with ease of return implied) comes next, and the cost to deliver or collect a product becomes one of the most relevant associations any retailer will have to profitability and customer satisfaction,” Charness said. Assess your supply chain and fulfillment capabilities, and benchmark them to competitors and best-in-class companies.
  • Post-sales service, resales and loyalty. “Your conversation with your customer doesn’t end with the shipment,” Charness Consider what else you can say to or advise your customer on to keep the relationship alive and productive.
  • Personalization everywhere. “Put yourself in your shopper’s shoes, and play back the conversations you’ve had during the entire shopper journey,” Charness said. “Was it always relevant, interesting and useful? Or did you communicate with mass marketing techniques, treating everyone the same?” Develop a strategy for using AI and machine learning across “the end-to-end interaction chain” with customers to enhance personalized service.

The future of e-commerce

Many businesses have been on their e-commerce journey for years, adapting business processes to the customer predilection for digital transactions. However, few were well prepared for the rapid and wholesale shift to digital transactions driven by the pandemic. A record-breaking 11,100-plus stores closed in the U.S. last year, and 40 major retailers filed for Chapter 11 bankruptcy protection, according to CoStar Group, a collector of retail real estate data. More stores are expected to shutter in the upcoming years, with some analysts predicting 100,000 stores — mostly apparel retailers — could close by 2025.

Yet, despite their struggles and challenges, many organizations are on their way to success. The awareness that challenges must be faced and addressed indicates that organizations understand that data-driven, personalized and secure customer transactions are the future.

How these transactions happen — whether online, via a mobile device, through some combination of digital and physical channels or by some augmented reality lens not yet imagined — will depend on circumstances and customer preferences, but they will increasingly involve digital technologies.

Indeed, Gartner has advocated replacing the term e-commerce with digital commerce to better reflect the convergence of all the digital systems that go into transactions today.

As customers increasingly decide that the frictionless experiences they have when buying online from leading digital vendors are the norm, the semantic distinction between e-commerce or digital commerce or any other kind of buying and selling transaction will disappear.

“When it’s all said and done,” Gillespie said, “we’ll just call it commerce.”

Coinsmart. Beste Bitcoin-Börse in Europa
Source: https://www.fintechnews.org/the-future-of-e-commerce-trends-tips-traps-to-avoid-2/

Continue Reading

Big Data

If you did not already know

Avatar

Published

on

Metalog Distribution google


In economics, business, engineering, science and other fields, continuous uncertainties frequently arise that are not easily- or well-characterized by previously-named continuous probability distributions. Frequently, there is data available from measurements, assessments, derivations, simulations or other sources that characterize the range of an uncertainty. But the underlying process that generated this data is either unknown or fails to lend itself to convenient derivation of equations that appropriately characterize the probability density (PDF), cumulative (CDF) or quantile distribution functions. The metalog distributions are a family of continuous univariate probability distributions that directly address this need. They can be used in most any situation in which CDF data is known and a flexible, simple, and easy-to-use continuous probability distribution is needed to represent that data. Consider their uses and benefits. Also consider their applications over a wide range of fields and data sources. …

Long Short-Term Attention (LSTA) google


Egocentric activity recognition is one of the most challenging tasks in video analysis. It requires a fine-grained discrimination of small objects and their manipulation. While some methods base on strong supervision and attention mechanisms, they are either annotation consuming or do not take spatio-temporal patterns into account. In this paper we propose LSTA as a mechanism to focus on features from spatial relevant parts while attention is being tracked smoothly across the video sequence. We demonstrate the effectiveness of LSTA on egocentric activity recognition with an end-to-end trainable two-stream architecture, achieving state of the art performance on four standard benchmarks. …

Gaussian Process Posterior Sampling Reinforcement Learning (GPPSTD) google


Efficient Reinforcement Learning usually takes advantage of demonstration or good exploration strategy. By applying posterior sampling in model-free RL under the hypothesis of GP, we propose Gaussian Process Posterior Sampling Reinforcement Learning(GPPSTD) algorithm in continuous state space, giving theoretical justifications and empirical results. We also provide theoretical and empirical results that various demonstration could lower expected uncertainty and benefit posterior sampling exploration. In this way, we combined the demonstration and exploration process together to achieve a more efficient reinforcement learning. …

SkinnerDB google


SkinnerDB is designed from the ground up for reliable join ordering. It maintains no data statistics and uses no cost or cardinality models. Instead, it uses reinforcement learning to learn optimal join orders on the fly, during the execution of the current query. To that purpose, we divide the execution of a query into many small time slices. Different join orders are tried in different time slices. We merge result tuples generated according to different join orders until a complete result is obtained. By measuring execution progress per time slice, we identify promising join orders as execution proceeds. Along with SkinnerDB, we introduce a new quality criterion for query execution strategies. We compare expected execution cost against execution cost for an optimal join order. SkinnerDB features multiple execution strategies that are optimized for that criterion. Some of them can be executed on top of existing database systems. For maximal performance, we introduce a customized execution engine, facilitating fast join order switching via specialized multi-way join algorithms and tuple representations. We experimentally compare SkinnerDB’s performance against various baselines, including MonetDB, Postgres, and adaptive processing methods. We consider various benchmarks, including the join order benchmark and TPC-H variants with user-defined functions. Overall, the overheads of reliable join ordering are negligible compared to the performance impact of the occasional, catastrophic join order choice. …

Coinsmart. Beste Bitcoin-Börse in Europa
Source: https://analytixon.com/2021/05/08/if-you-did-not-already-know-1391/

Continue Reading
PR Newswire2 days ago

Polystyrene Foam Market worth $32.2 billion by 2026 – Exclusive Report by MarketsandMarkets™

Blockchain5 days ago

The Reason for Ethereum’s Recent Rally to ATH According to Changpeng Zhao

Blockchain4 days ago

Chiliz Price Prediction 2021-2025: $1.76 By the End of 2025

Blockchain5 days ago

Mining Bitcoin: How to Mine Bitcoin

Blockchain5 days ago

Mining Bitcoin: How to Mine Bitcoin

PR Newswire4 days ago

Teamsters Lead Historic Defeat of CEO Pay at Marathon Petroleum

Blockchain5 days ago

Amid XRP lawsuit, Ripple appoints former US Treasurer to its board, and names new CFO

Blockchain5 days ago

Mining Bitcoin: How to Mine Bitcoin

Esports5 days ago

TFT 11.9 B-patch nerfs Mordekaiser and LeBlanc

AR/VR3 days ago

Apple is giving a laser company that builds some of its AR tech $410 million

Blockchain4 days ago

Galaxy Digital Set To Buy BitGo for $1.2 Billion

Blockchain5 days ago

‘DeFi may lead to a paradigm shift’ says Federal Reserve Bank paper

Blockchain2 days ago

Launch of Crypto Trading Team by Goldman Sachs

Esports5 days ago

Umbreon VMAX, Glaceon VMAX, and Eeveelution Trainer items revealed for Pokémon OCG set Eevee Heroes

Private Equity4 days ago

Beyond the fanfare and SEC warnings, SPACs are here to stay

Automotive4 days ago

Brembo Debuts Light-Up LED Brake Calipers

Aviation1 day ago

What Happened To Lufthansa’s Boeing 707 Aircraft?

Esports5 days ago

When does Destiny 2 Season of the Splicer start and end?

Payments3 days ago

G20 TechSprint Initiative invites firm to tackle green finance

Blockchain5 days ago

New York bill proposes to ban crypto mining for 3 years over carbon concerns

Trending