Take A Deeper Look at Deep Learning

Image: Pixabay

1. Deep learning is useful only for complex problems.

2. Deep learning solves the feature engineering problem.

3. People fear the ‘black box.’

4. Deep learning technology has the ability to transfer learning

5. Deep learning is good at logistic regression.

6. Most deep learning projects involve gradient descent.

7. Deep learning comes in many forms.

8. Convolutional neural networks are one of most common types of neural networks.

9. Deep learning is now better than humans at many tasks.

https://www.informationweek.com/big-data/ai-machine-learning/take-a-deeper-look-at-deep-learning/d/d-id/1333903

3 things we learned from Facebook’s AI chief about the future of artificial intelligence

Facebook’s chief AI scientist Yann LeCun

“The International Data Corporation indicates global spending on AI systems is expected to hit $US77.6 billion in 2022, more than tripling the $US24 billion forecast for 2018.

A key element in advancing the field of artificial intelligence, particularly when it comes to deep learning, will be ensuring that there’s hardware capable of supporting it.

  1. Machines have to get much better at power consumption in order for AI to improve.
  2. We’ll continue seeing AI advancements in smartphones in the near term before improvements appear elsewhere.
  3. Giving machines “common sense” will be a big focus for AI research in the next decade.”

https://www.google.com/amp/s/amp.businessinsider.com/facebook-artificial-intelligence-yann-lecun-2019-2

The State of Machine Intelligence 3.0

Looking Forward

“We see all this activity only continuing to accelerate. The world will give us more open sourced and commercially available machine intelligence building blocks; there will be more data; there will be more people interested in learning these methods; and there will always be problems worth solving. We still need ways of explaining the difference between machine intelligence and traditional software, and we’re working on that. The value of code is different from data, but what about the value of the model that code improves based on that data?

Once we understand machine intelligence deeply, we might look back on the era of traditional software and think it was just a prologue to what’s happening now. We look forward to seeing what the next year brings.”

https://www.oreilly.com/library/view/artificial-intelligence-now/9781492049210/ch01.html

The Deep Learning Framework Backed By Facebook Is Getting Industry’s Attention

Neural NetworksSOURCE: PIXABAY

“When it comes to deep learning frameworks, TensorFlow is one of the most preferred toolkits. However, one framework that is fast becoming the favorite of developers and data scientists is PyTorch.

PyTorch is an open source project from Facebook which is used extensively within the company.

PyTorch focuses on simplicity and accessibility. It can be used by a diverse set of users ranging from researchers to academicians to a developer. PyTorch uses a technique known as dynamic computation that makes it easy to train neural networks. TensorFlow is based on static computation that executes the code only after the graph of operations is generated.”

https://www.google.com/amp/s/www.forbes.com/sites/janakirammsv/2019/02/11/the-deep-learning-framework-backed-by-facebook-is-getting-industrys-attention/amp/

Zillow awards $1 million to team that reduced home valuation algorithm error to below 4%

  • “The Seattle company today announced that team ChaNJestimate — whose members include Chahhou Mohamed, Jordan Meyer, and Nima Shahbazi, hailing from Morocco, the U.S., and Canada, respectively — will take home the $1 million prize for a model that bested the Zillow “Zestimate” benchmark by approximately 13 percent. (The Zestimate’s nationwide error rate is 4.5 percent; the team’s work pushes it to below 4 percent.)
  • To achieve this new level of accuracy, team ChaNJestimate leveraged deep neural networks — layers of mathematical models modeled after neurons in the brain — and other machine learning techniques to “directly” estimate home values
  • Currently, the Zestimate is within $10,000 of a given home’s sale price, and Zillow expects the improvements could bring it $1,300 closer to the actual price.”

https://www.google.com/amp/s/venturebeat.com/2019/01/30/zillow-awards-1-million-to-team-that-reduced-home-valuation-algorithm-error-to-below-4/amp/

We analyzed 16,625 papers to figure out where AI is headed next

The rise of reinforcement learning

“In the few years since the rise of deep learning, our analysis reveals, a third and final shift has taken place in AI research.

As well as the different techniques in machine learning, there are three different types: supervised, unsupervised, and reinforcement learning. Supervised learning, which involves feeding a machine labeled data, is the most commonly used and also has the most practical applications by far. In the last few years, however, reinforcement learning, which mimics the process of training animals through punishments and rewards, has seen a rapid uptick of mentions in paper abstracts.

The idea isn’t new, but for many decades it didn’t really work. “The supervised-learning people would make fun of the reinforcement-learning people,” Domingos says. But, just as with deep learning, one pivotal moment suddenly placed it on the map.

That moment came in October 2015, when DeepMind’s AlphaGo, trained with reinforcement learning, defeated the world champion in the ancient game of Go. The effect on the research community was immediate. 

The next decade

Our analysis provides only the most recent snapshot of the competition among ideas that characterizes AI research. But it illustrates the fickleness of the quest to duplicate intelligence. “The key thing to realize is that nobody knows how to solve this problem,” Domingos says.

Many of the techniques used in the last 25 years originated at around the same time, in the 1950s, and have fallen in and out of favor with the challenges and successes of each decade. Neural networks, for example, peaked in the ’60s and briefly in the ’80s but nearly died before regaining their current popularity through deep learning.

Every decade, in other words, has essentially seen the reign of a different technique: neural networks in the late ’50s and ’60s, various symbolic approaches in the ’70s, knowledge-based systems in the ’80s, Bayesian networks in the ’90s, support vector machines in the ’00s, and neural networks again in the ’10s.

The 2020s should be no different, says Domingos, meaning the era of deep learning may soon come to an end. But characteristically, the research community has competing ideas about what will come next—whether an older technique will regain favor or whether the field will create an entirely new paradigm.”

https://www.technologyreview.com/s/612768/we-analyzed-16625-papers-to-figure-out-where-ai-is-headed-next/

Become an AI Company in 90 Days

“Researchers are continually publishing effective AI models. Hardware is a commodity. But build the highest-quality proprietary dataset and you’ll crush the competition every time.”

I had a day off and the electrician took down my power for the day so I had some much anticipated time on my hands. And just as any other geek would do…I used the time to tear through an unsolicited but gratis book I received in the mail over the holidays.

The book titled, “Become An AI Company in 90 Days,” was written by Kevin Dewalt and produced by Russ Rands.

The book is broken into 4 parts including:

  • AI Fundamentals
  • Discovering AI Opportunities
  • Building a Winning Strategy
  • Launching Your First AI Product

Here’s the scoop:

    Author is legitimate and knows his stuff (Stanford AI experience)
    He has clearly worked with some large corporations based on some of his anecdotes (policy and cross functional alignment will be some of the toughest challenges)
    The instructions are specific and practical (top of line system for under $5K and reasonable data set size…good rule of thumb is 50,000 labeled examples)
    Book length of 132 pages is easily consumable in an afternoon
    Content is well written and understandable without an advanced degree in mathematics
    Includes examples that help bring the material to life (automated claims processing for the AI Canvas tool)

My Review: It’s a solid “A” and I recommend you check it out.

Get a free copy: https://prolego.io/book.html

Note: I do not have any personal or professional connection to the authors, and I do not have any financial interest in the company or the book

An executive’s guide to AI – McKinsey

  • One of the highest rated McKinsey articles of the year is titled, “An executive’s guide to AI.” It’s a quick overview including key elements and terminology.
  • Artificial Intelligence
    • AI is typically defined as the ability of a machine to perform cognitive functions we associate with human minds, such as perceiving, reasoning, learning, and problem solving.
    • Timeline
  • Machine Learning
    • Machine- learning algorithms detect patterns and learn how to make predictions and recommendations by processing data and experiences, rather than by receiving explicit programming instruction.

    Types of analytics

    • Descriptive, predictive, prescriptive
  • Major Types
    • Supervised, unsupervised, reinforcement

    Deep Learning

    • Deep learning is a type of machine learning that can process a wider range of data resources, requires less data preprocessing by humans, and can often produce more accurate results than traditional machine-learning approaches.
  • Major Models
    • Convolutional neural network CNN
    • Recurrent neural network

    Business Cases

      Predict Call Center Volume
    • Predict power usage in an electrical grid distribution
    • Detect fraudulent activity for credit cards
    • Simple low-cost image classification
    • Forecast product demand and inventory
    • Predict the price of cars
    • Predict probability of patient joining healthcare
    • Predict prices that will be paid for a product

    https://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/an-executives-guide-to-aihttps://www.mckinsey.com/business-functions/mckinsey-analytics/our-insights/an-executives-guide-to-ai

    Amazon has an AI vision and Rajeev Rastogi is shaping it

    “Rastogi is the director of machine learning at Amazon India and has been ranked among one of the top 10 data scientists in India. The e-commerce giant is looking at putting artificial intelligence and ML at the heart of everything it does. According to Rastogi, the problems that AI and ML need to spar within India are very different from the Amazon universe in the rest of the world, especially because the data available is not of high quality.

    Rastogi had been dabbling with data even earlier – he holds a PhD in computer science from the University of Austin at Texas and more than 50 patents – and Bell Labs India was the springboard into a career of solving problems using data. Yahoo! Labs came after the Bell Labs gig before Amazon convinced to switch over.”

    https://factordaily.com/amazon-has-an-ai-vision-and-rajeev-rastogi-is-shaping-it/