Connect with us

Published

on

Deep neural networks will move past their shortcomings without help from symbolic artificial intelligence, three pioneers of deep learning argue in a paper published in the July issue of the Communications of the ACM journal.

In their paper, Yoshua Bengio, Geoffrey Hinton, and Yann LeCun, recipients of the 2018 Turing Award, explain the current challenges of deep learning and how it differs from learning in humans and animals. They also explore recent advances in the field that might provide blueprints for the future directions for research in deep learning.

Titled “Deep Learning for AI,” the paper envisions a future in which deep learning models can learn with little or no help from humans, are flexible to changes in their environment, and can solve a wide range of reflexive and cognitive problems.

The challenges of deep learning

Deep learning is often compared to the brains of humans and animals. However, the past years have proven that artificial neural networks, the main component used in deep learning models, lack the efficiency, flexibility, and versatility of their biological counterparts.

In their paper, Bengio, Hinton, and LeCun acknowledge these shortcomings. “Supervised learning, while successful in a wide variety of tasks, typically requires a large amount of human-labeled data. Similarly, when reinforcement learning is based only on rewards, it requires a very large number of interactions,” they write.

Supervised learning is a popular subset of machine learning algorithms, in which a model is presented with labeled examples, such as a list of images and their corresponding content. The model is trained to find recurring patterns in examples that have similar labels. It then uses the learned patterns to associate new examples with the right labels. Supervised learning is especially useful for problems where labeled examples are abundantly available.

Reinforcement learning is another branch of machine learning, in which an “agent” learns to maximize “rewards” in an environment. An environment can be as simple as a tic-tac-toe board in which an AI player is rewarded for lining up three Xs or Os, or as complex as an urban setting in which a self-driving car is rewarded for avoiding collisions, obeying traffic rules, and reaching its destination. The agent starts by taking random actions. As it receives feedback from its environment, it finds sequences of actions that provide better rewards.

In both cases, as the scientists acknowledge, machine learning models require huge labor. Labeled datasets are hard to come by, especially in specialized fields that don’t have public, open-source datasets, which means they need the hard and expensive labor of human annotators. And complicated reinforcement learning models require massive computational resources to run a vast number of training episodes, which makes them available to a few, very wealthy AI labs and tech companies.

Bengio, Hinton, and LeCun also acknowledge that current deep learning systems are still limited in the scope of problems they can solve. They perform well on specialized tasks but “are often brittle outside of the narrow domain they have been trained on.” Often, slight changes such as a few modified pixels in an image or a very slight alteration of rules in the environment can cause deep learning systems to go astray.

The brittleness of deep learning systems is largely due to machine learning models being based on the “independent and identically distributed” (i.i.d.) assumption, which supposes that real-world data has the same distribution as the training data. i.i.d also assumes that observations do not affect each other (e.g., coin or die tosses are independent of each other).

“From the early days, theoreticians of machine learning have focused on the iid assumption… Unfortunately, this is not a realistic assumption in the real world,” the scientists write.

Real-world settings are constantly changing due to different factors, many of which are virtually impossible to represent without causal models. Intelligent agents must constantly observe and learn from their environment and other agents, and they must adapt their behavior to changes.

“[T]he performance of today’s best AI systems tends to take a hit when they go from the lab to the field,” the scientists write.

The i.i.d. assumption becomes even more fragile when applied to fields such as computer vision and natural language processing, where the agent must deal with high-entropy environments. Currently, many researchers and companies try to overcome the limits of deep learning by training neural networks on more data, hoping that larger datasets will cover a wider distribution and reduce the chances of failure in the real world.

Deep learning vs hybrid AI

The ultimate goal of AI scientists is to replicate the kind of general intelligence humans have. And we know that humans don’t suffer from the problems of current deep learning systems.

“Humans and animals seem to be able to learn massive amounts of background knowledge about the world, largely by observation, in a task-independent manner,” Bengio, Hinton, and LeCun write in their paper. “This knowledge underpins common sense and allows humans to learn complex tasks, such as driving, with just a few hours of practice.”

Elsewhere in the paper, the scientists note, “[H]umans can generalize in a way that is different and more powerful than ordinary iid generalization: we can correctly interpret novel combinations of existing concepts, even if those combinations are extremely unlikely under our training distribution, so long as they respect high-level syntactic and semantic patterns we have already learned.”

Scientists provide various solutions to close the gap between AI and human intelligence. One approach that has been widely discussed in the past few years is hybrid artificial intelligence that combines neural networks with classical symbolic systems. Symbol manipulation is a very important part of humans’ ability to reason about the world. It is also one of the great challenges of deep learning systems.

Bengio, Hinton, and LeCun do not believe in mixing neural networks and symbolic AI. In a video that accompanies the ACM paper, Bengio says, “There are some who believe that there are problems that neural networks just cannot resolve and that we have to resort to the classical AI, symbolic approach. But our work suggests otherwise.”

The deep learning pioneers believe that better neural network architectures will eventually lead to all aspects of human and animal intelligence, including symbol manipulation, reasoning, causal inference, and common sense.

Promising advances in deep learning

In their paper, Bengio, Hinton, and LeCun highlight recent advances in deep learning that have helped make progress in some of the fields where deep learning struggles.

One example is the Transformer, a neural network architecture that has been at the heart of language models such as OpenAI’s GPT-3 and Google’s Meena. One of the benefits of Transformers is their capability to learn without the need for labeled data. Transformers can develop representations through unsupervised learning, and then they can apply those representations to fill in the blanks on incomplete sentences or generate coherent text after receiving a prompt.

More recently, researchers have shown that Transformers can be applied to computer vision tasks as well. When combined with convolutional neural networks, transformers can predict the content of masked regions.

A more promising technique is contrastive learning, which tries to find vector representations of missing regions instead of predicting exact pixel values. This is an intriguing approach and seems to be much closer to what the human mind does. When we see an image such as the one below, we might not be able to visualize a photo-realistic depiction of the missing parts, but our mind can come up with a high-level representation of what might go in those masked regions (e.g., doors, windows, etc.). (My own observation: This can tie in well with other research in the field aiming to align vector representations in neural networks with real-world concepts.)

The push for making neural networks less reliant on human-labeled data fits in the discussion of self-supervised learning, a concept that LeCun is working on.

Can you guess what is behind the grey boxes in the above image?

The paper also touches upon “system 2 deep learning,” a term borrowed from Nobel laureate psychologist Daniel Kahneman. System 2 accounts for the functions of the brain that require conscious thinking, which include symbol manipulation, reasoning, multi-step planning, and solving complex mathematical problems. System 2 deep learning is still in its early stages, but if it becomes a reality, it can solve some of the key problems of neural networks, including out-of-distribution generalization, causal inference, robust transfer learning, and symbol manipulation.

The scientists also support work on “Neural networks that assign intrinsic frames of reference to objects and their parts and recognize objects by using the geometric relationships.” This is a reference to “capsule networks,” an area of research Hinton has focused on in the past few years. Capsule networks aim to upgrade neural networks from detecting features in images to detecting objects, their physical properties, and their hierarchical relations with each other. Capsule networks can provide deep learning with “intuitive physics,” a capability that allows humans and animals to understand three-dimensional environments.

“There’s still a long way to go in terms of our understanding of how to make neural networks really effective. And we expect there to be radically new ideas,” Hinton told ACM.

This article was originally published by Ben Dickson on TechTalks, a publication that examines trends in technology, how they affect the way we live and do business, and the problems they solve. But we also discuss the evil side of technology, the darker implications of new tech, and what we need to look out for. You can read the original article here.

Advertisements



Naabiae Nenu-B is a Medical Health Student and an SEO Specialist dedicated to flushing the web off fake news and scam scandals. He aims at being "Africa's Best Leak and Review Blogger" and that's the unwavering stand of Xycinews Media.

Advertisement
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Technology

Amazon’s best MacBook Pro deals of 2021 are available right now

Published

on

Are you looking for the fastest and most responsive experience possible on a laptop computer? You don’t need to spend a small fortune on a custom gaming machine with a gazillion gigabits of RAM. All you need is a new Apple notebook with Apple’s incredible M1 chip. And prices actually start much lower than you might think! That’s especially true for Apple’s Pro-grade M1 laptop. As a matter of fact, Amazon’s best MacBook Pro deals of 2021 are happening right now!

2020 Apple MacBook Pro with Apple M1 Chip (256GB) List Price:$1,299.00 Price:$1,099.99 You Save:$199.01 (15%) Buy Now Available from Amazon, BGR may receive a commission

The latest-generation M1 MacBook Air is Apple’s most affordable laptop you can get with the new M1 chipset. As if you don’t already know, Apple’s M1 is the SoC that’s making all kinds of waves right now. This entry-level Apple laptop is still powerful enough to computer crush rival Windows laptops that cost twice as much. And that’s especially true while Amazon is discounting the MacBook Air by up to $150.

Best MacBook Pro deals

Apple’s new Air laptop is lightning-fast indeed, but there’s another M1-powered Apple notebook that’s even more powerful. If you want the best of the best, you’re likely eying a brand new Apple M1 MacBook Pro instead of the Air. If that’s the case, we have some fantastic news: Now is the time to pull the trigger because Amazon is offering all-time low prices on both the 256GB model and the 512GB version, with $200 discounts on either size!

Searching for the fastest and most powerful Apple computers ever made? You now have four models to choose from. The new Mac mini is one, plus the new MacBook Air, and 13-inch MacBook Pro. And now, the new M1 iMac has finally been released (plus, it’s $50 off at Amazon!). These computers are all insanely powerful, and they’re all very new to Apple’s lineup. Despite how new they are though, Amazon already has some remarkably deep discounts.

Apple’s new Mac mini is the only M1-powered desktop that isn’t an all-in-one model. It’s so affordable, especially considering the fact that it crushes Windows PCs that cost twice as much. The same can be said of the MacBook Air, which is also on sale with discounts up to $150. But it should go without saying that the most exciting new M1-powered computer is the M1 MacBook Pro. And Amazon just slashed this incredible laptop to its lowest price ever.

The M1 MacBook Pro was between $50 and $60 off ahead of the holidays last year. Procrastination doesn’t often off, but this time around you’ll be very glad that you missed those deals. Head over to Amazon today and you’ll find that the 13-inch MacBook Pro with 256GB of SSD storage is $200 off. And if you want to bump your storage space up to 512GB you’ll still save a whopping $200!

Both of those deals are all-time low prices for Apple’s new M1 MacBook Pro models. But the bad news is we have no idea how much longer Amazon’s awesome sale will last. In other words, hurry up or you could miss out!


M1 MacBook Pro

  • Apple’s M1 chip is a revolution in the personal computing space, offering unrivaled CPU and GPU performance
  • Instant wake means no more waiting for your laptop to load
  • The 8-core CPU delivers up to 2.8x faster performance, while the 8-core GPU brings up to 5x faster graphics
  • The M1 is also Apple’s most efficient chipset ever, enabling up to 20 hours of battery life on a single charge
  • The dedicated 16-core Neural Engine enables advanced machine learning capabilities
  • There has never been anything like this before on comparable notebook computers

2020 Apple MacBook Pro with Apple M1 Chip (256GB) List Price:$1,299.00 Price:$1,099.99 You Save:$199.01 (15%) Buy Now Available from Amazon, BGR may receive a commission
2020 Apple MacBook Pro with Apple M1 Chip (512GB) List Price:$1,499.00 Price:$1,299.99 You Save:$199.01 (13%) Buy Now Available from Amazon, BGR may receive a commission


M1 MacBook Air

  • Powered by Apple’s M1 chip, the first M series chip designed entirely by Apple to optimize performance and efficiency
  • Apple’s 8-core CPU delivers up to 2.8x faster performance and the 8-core GPU powers up to 5x faster graphics
  • Up to 18 hours of battery life on a single charge — leave your bulky power supply at home
  • Instant wake enabled by the M1 means your MacBook is ready to go as soon as you open it

2020 Apple MacBook Air Laptop with Apple M1 Chip (256GB) List Price:$999.00 Price:$899.00 You Save:$100.00 (10%) Buy Now Available from Amazon, BGR may receive a commission
2020 Apple MacBook Air Laptop with Apple M1 Chip (512GB) List Price:$1,249.00 Price:$1,099.99 You Save:$149.01 (12%) Buy Now Available from Amazon, BGR may receive a commission


Go here to see this month’s best deals on Amazon!

Advertisements



Continue Reading

Technology

HBO Max free episodes are available right now to anyone

Published

on

Over the last few years, the number of streaming services on the market has increased at an alarming pace. Netflix, Hulu, Amazon Prime Video, Peacock, Paramount+, Disney+, HBO Max, and more are all competing for your time and money. But there are only so many hours in a day and dollars in our bank accounts. It’s a fierce competition, which is why every service is constantly finding new ways to reel in subscribers. To that point, HBO announced this week that episodes from some of the biggest shows on HBO Max will be free to watch for non-subscribers.

Today’s Top Deal Fire TV Stick 4K just got a rare 20% discount — don’t miss out! List Price:$49.99 Price:$39.99 You Save:$10.00 (20%) Buy Now Available from Amazon, BGR may receive a commission Available from Amazon BGR may receive a commission

If you’re curious about the HBO Max library, this is a great way to get a sneak peek before spending any money. HBO says the collection will initially include 13 episodes of HBO and Max Originals. New titles will rotate in and out of the collection periodically. Here are all of the pilot episodes that HBO says you can watch for free without a subscription:

With that said, when I visit the HBO Max free episodes hub, I see far more than what is listed above. Beyond those shows, I also see The Nevers, Mare of Easttown, The Undoing, Insecure, I May Destroy You, and Gossip Girl, to name a few. Perhaps those were already part of a separate promotion.

How to watch HBO Max free episodes

There are a number of ways to watch the free episodes on HBO Max. In addition to clicking on the link above in your browser, you can also access the free episodes on all of HBO Max’s device partners. This includes Fire TV, Android TV, Android, Apple TV, Chromecast, iOS, LG, PlayStation, Roku, Samsung, and Xbox. Basically, if you have the app, you can watch the free episodes.

Signing up for HBO Max

If the company’s gambit works, these free episodes might convince you to subscribe to HBO Max. If you’re a movie buff, you might want to take advantage of the ad-free tier for $14.99 a month. For the rest of the year, Warner Bros. Pictures movies will premiere in theaters and on HBO Max simultaneously. If you’re not interested in watching The Matrix 4 or Dune at home, you might prefer the cheaper tier, which costs $9.99 a month, but includes ads.

Today’s Top Deal Fire TV Stick 4K just got a rare 20% discount — don’t miss out! List Price:$49.99 Price:$39.99 You Save:$10.00 (20%) Buy Now Available from Amazon, BGR may receive a commission Available from Amazon BGR may receive a commission

Advertisements



Continue Reading

Technology

Windows 365 trials are now locked because everyone’s swarming to try it

Published

on

Microsoft brought Windows 10 to the iPad a few weeks ago, in what seemed like an extraordinary feat. However, you don’t actually get to install Windows on the iPad, and Apple will never allow that. But you can run the full Windows 10 experience on any internet-connected device that supports an internet browser. That’s because Windows 365 runs in the cloud, allowing businesses to offer their customers virtual Cloud PCs. Admins can customize these cloud computers in real-time, offering the kind of upgrades and downgrades that aren’t possible with real computers. Microsoft made Windows 365 available to businesses earlier this week, and it didn’t take long for the new Windows experience to sell out. Microsoft confirmed as much on social media, noting that the Windows 365 trial is locked for the time being.

Today’s Top Deal The best Alexa smart plugs on Amazon are somehow down to just $5 each! List Price:$24.99 Price:$19.99 You Save:$5.00 (20%) Buy Now Available from Amazon, BGR may receive a commission Available from Amazon BGR may receive a commission


It might seem impossible for a cloud service to sell out, but Microsoft ran out of cloud capacity to meet the massive demand. Brand new devices routinely sell out as companies hurry to mass-produce more units. But with a device in the cloud, the buyer doesn’t interact with a physical device. Microsoft could theoretically offer infinite Windows 365 virtual computers. But, in practice, that’s impossible.

Microsoft needs to have the cloud infrastructure in place to support all these concurrent Windows 365 sessions. And the company has run out of capacity less than two days after making Windows 365 available to businesses.

How Windows 365 works

Users added to a Windows 365 trial subscription will consume the virtual PC resources allotted to them. Admins can configure only certain key aspects of a Cloud PC, including CPU, RAM, and storage. They can add and remove resources as needed. They can remove people as well.

That’s a great type of Windows product that some businesses might find very useful. Rather than buying and maintaining new hardware for employees, some companies might want to deploy fleets of virtual Cloud PCs. Each employee gets their unique Cloud PC that’s hooked up to fast internet. And they can access their Windows 365 PC from any device that can connect to the internet. Hence, running the full Windows 10 experience on iPad is now possible.

Best of all, the user’s hardware will not impact Windows 365 performance. The Cloud PC experience is streamed to any internet-connected device. It’s the internet connection that matters here, not how powerful the user’s personal computer is.

Microsoft paused the trial

Microsoft did not disclose full pricing details for Windows 365 in mid-July when it announced the service. But the company opened the trial on August 3rd, revealing that Cloud PC rentals start at $20 per user per month. The cheapest Cloud PC features one CPU, 2GB of RAM, and 64GB of storage. The price goes up to $24 per month if you don’t own a Windows 10 Professional license.

That’s a great starting price, but it’s only available to businesses. Regular users who dream of a full-fledged Windows PC in the cloud can’t get on Windows 365. The best part of the service is that businesses get to try it for free for two months. That helps explain why Microsoft sold out all its Windows 365 trial “units.”

Microsoft confirmed everything on Twitter. Businesses that weren’t sure whether or not to get on the Windows 365 trial a few days ago will now have to wait. They can sign up to be notified when cloud resources become available.

Today’s Top Deal Apple AirPods 2 are down to Amazon’s lowest price of 2021 List Price:$159.00 Price:$119.00 You Save:$40.00 (25%) Buy Now Available from Amazon, BGR may receive a commission Available from Amazon BGR may receive a commission

Advertisements



Continue Reading

Trending

  • bitcoinBitcoin (BTC) $ 38,939.00
  • ethereumEthereum (ETH) $ 2,687.61
  • tetherTether (USDT) $ 1.00
  • binance-coinBinance Coin (BNB) $ 330.52
  • cardanoCardano (ADA) $ 1.37
  • xrpXRP (XRP) $ 0.722550
  • usd-coinUSD Coin (USDC) $ 1.00
  • dogecoinDogecoin (DOGE) $ 0.199105
  • polkadotPolkadot (DOT) $ 18.98
  • uniswapUniswap (UNI) $ 23.23
  • binance-usdBinance USD (BUSD) $ 1.00
  • chainlinkChainlink (LINK) $ 23.69
  • bitcoin-cashBitcoin Cash (BCH) $ 534.96
  • solanaSolana (SOL) $ 36.83
  • litecoinLitecoin (LTC) $ 139.65
  • wrapped-bitcoinWrapped Bitcoin (WBTC) $ 38,971.00
  • matic-networkPolygon (MATIC) $ 1.04
  • ethereum-classicEthereum Classic (ETC) $ 50.82
  • stellarStellar (XLM) $ 0.274902
  • theta-tokenTheta Network (THETA) $ 6.09
  • terra-lunaTerra (LUNA) $ 14.63
  • vechainVeChain (VET) $ 0.088028
  • internet-computerInternet Computer (ICP) $ 40.00
  • daiDai (DAI) $ 0.999577
  • filecoinFilecoin (FIL) $ 56.35
  • compound-usd-coincUSDC (CUSDC) $ 0.022146
  • tronTRON (TRX) $ 0.070134
  • okbOKB (OKB) $ 18.52
  • aaveAave (AAVE) $ 361.02
  • moneroMonero (XMR) $ 244.08
  • cdaicDAI (CDAI) $ 0.021600
  • compound-ethercETH (CETH) $ 53.83
  • ftx-tokenFTX Token (FTT) $ 37.18
  • eosEOS (EOS) $ 4.08
  • cosmosCosmos (ATOM) $ 12.85
  • pancakeswap-tokenPancakeSwap (CAKE) $ 16.98
  • crypto-com-chainCrypto.com Coin (CRO) $ 0.122547
  • the-graphThe Graph (GRT) $ 0.657408
  • shiba-inuShiba Inu (SHIB) $ 0.000006
  • neoNEO (NEO) $ 43.01
  • amp-tokenAmp (AMP) $ 0.061081
  • leo-tokenLEO Token (LEO) $ 3.00
  • tezosTezos (XTZ) $ 3.25
  • algorandAlgorand (ALGO) $ 0.844094
  • klay-tokenKlaytn (KLAY) $ 1.07
  • makerMaker (MKR) $ 2,953.66
  • bitcoin-svBitcoin SV (BSV) $ 137.65
  • celsius-degree-tokenCelsius Network (CEL) $ 5.86
  • compound-governance-tokenCompound (COMP) $ 459.07
  • iotaIOTA (MIOTA) $ 0.856133
error: Content is protected !!