Machine learning, shall we?

Why should we use machine learning and what are reasons why we should restrict its use

Uwe Sterr http://uwesterr.de
01-21-2020

Table of Contents


Before we answer the question Machine learning, shall we? lets look into what machine learning is

What is machine learning?

Machine learning is a sub domain of artificial intelligence and is defined:

What is intelligence?

To discuss the question of what is artificial intelligence, the first step is to define what intelligence is.

A group of 52 psychology researchers published in (Gottfredson 1997) the following definition:

A very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test-taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings—“catching on,” “making sense” of things, or “figuring out” what to do.

Alfred Binet, a french psychologist who invented first practical IQ test defined in 1905 (Binet and Simon 1916):

Judgment, otherwise called “good sense”, “practical sense”, “initiative”, the faculty of adapting one’s self to circumstances

And Albert Einstein said

The measure of intelligence is the ability to change.

Tegmark’s summarizes the situation in (Tegmark 2017)

There’s no agreement on what intelligence is even among intelligent intelligence researchers! So there’s clearly no undisputed “correct” definition of intelligence.

Definition of artificial intelligence sub domains

Even though there is no undisputed definition of intelligence there is a undisputed definition of how machine learning is related to artificial intelligence

Agreement: Machine learning is a sub domain of artificial intelligence (AI)



The history of those fields goes back to the 1950’s



Is AI smarter than humans?

Which of the following questions can a computer answer better?

Thinking, fast and slow (Kahneman 2011)

In Daniel Kahneman’s Thinking, fast and slow (Kahneman 2011) there are plenty of surprising social psychology experiments, on page 166 the following question is posed:

A cab was involved in a hit-and-run accident at night. Two cab companies, the Green and the Blue, operate in the city. You are given the following data:
- 85% of the cabs in the city are Green and 15% are Blue.
- A witness identified the cab as Blue. The court tested the reliability of the witness under the circumstances that existed on the night of the accident and concluded that the witness correctly identified each one of the two colors 80% of the time and failed 20% of the time.

What is the probability that the cab involved in the accident was Blue rather than Green?6

Please cast your vote at https://pingo.coactum.de/1576787 or access Pingo webpage scanning QR code below:

Comparisons between AI and humans

Breast cancer detection

In a Google Health project8 the following results were achieved: Smiley face

In an independent study of six radiologists, the AI system outperformed all of the human readers. More on the study at https://www.nature.com/articles/s41586-019-1799-6

Working together: Lung cancer detection

With an estimated 160,000 deaths in 2018, lung cancer is the most common cause of cancer death in the United States

A study published in Nature medicine9 a team of members of Google AI and several hospitals reported

When prior computed tomography imaging was not available

ImageNet Large Scale Visual Recognition Challenge (ILSVRC)

The ImageNet Large Scale Visual Recognition Challenge10 (ILSVRC) (Russakovsky et al. 2015) evaluates algorithms for object recognition and image classification on a large scale.

Facts of ImageNet:11

The development of the results is shown in the graph below. The number of layers is a indication of model complexity



In 2017 the problem set to status “solved”
- 29 of 38 competing teams had an accuracy of more than 95%
- ImageNet stopped competition

AlphaGo Zero

Go is a strategy game invented 2500 years ago and has an estimated number of possible board configuration of 10¹⁷⁴ compared to chess which has is 10¹²º. A detailed description is given by DeepMind’s blog post “AlphaGo Zero: Starting from scratch”12

AlphaGo Zero is a version of DeepMind’s13 Go software AlphaGo

AlphaZero learned three games,

The capability progress of Alpha Zero during training is shown below

Figure from https://deepmind.com/blog/article/alphazero-shedding-new-light-grand-games-chess-shogi-and-go
Figure from https://deepmind.com/blog/article/alphazero-shedding-new-light-grand-games-chess-shogi-and-go

NOTE: EACH TRAINING STEP REPRESENTS 4,096 BOARD POSITIONS.

At the end of the training Alpha Zero achieved the following performance:

Figure from https://deepmind.com/blog/article/alphazero-shedding-new-light-grand-games-chess-shogi-and-go
Figure from https://deepmind.com/blog/article/alphazero-shedding-new-light-grand-games-chess-shogi-and-go

Implications are wider than just playing a game, as Garry Kasparov, a former world chess champion puts it:

The implications go far beyond my beloved chessboard… Not only do these self-taught expert machines perform incredibly well, but we can actually learn from the new knowledge they produce."

Bias

Models might end up biased, why is that?

How to avoid bias

Avoiding bias is harder than you might think

Human bias

Machine learning model can be biased for several reasons as shown above, how about humans?

Attacks

Especially image classification models have shown to be susceptible to attacks which leads to wrong classifications. This could lead to

How a attack can be performed is described by Goodfellow et al. in (Goodfellow, Shlens, and Szegedy 2014)

Figure from Image Credit: Goodfellow et al. (Goodfellow, Shlens, and Szegedy 2014))
Figure from Image Credit: Goodfellow et al. (Goodfellow, Shlens, and Szegedy 2014))

But what about attacks on human perception?

Which statement is correct?


Is this a picture of a real person?

Look at the picture below, is it a real person or an animation?

Figure from https://commons.wikimedia.org/wiki/File:Woman_1.jpg (Image Credit: Owlsmcgee [Public domain] )
Figure from https://commons.wikimedia.org/wiki/File:Woman_1.jpg (Image Credit: Owlsmcgee [Public domain] )

The image is create using a generative adversarial network (GAN), see below for the principle, for detailed description see https://medium.com/ai-society/gans-from-scratch-1-a-deep-introduction-with-code-in-pytorch-and-tensorflow-cb03cdcdba0f



Outlook

What will the future bring for society?

The saddest aspect of life right now is that science gathers knowledge faster than society gathers wisdom.

Isaac Asimov

Development of life

Tegmark in “Life 3.0: Being human in the age of artificial intelligence” (Tegmark 2017) p. 23. classifies life into three stages and shows the two existing stages of life and the third stage which might be ahead.



The three stages of life have overlapping skills, but only life 3.0 has all skills and is able to design its hardware and therefore might be able to have unlimited skills


When will superhuman AI come, and will it be good?

Several opinions about when and if superhuman AI will appear and if it will be a good thing or not exists. Those opinions can be grouped as shown in the following graph.

Luddite => A person opposed to new technology or ways of working

Please cast your vote at https://pingo.coactum.de/15767814

AI aftermath scenario

To be prepared we might want to ask yourselves:

  1. Do you want there to be superintelligence?
  2. Do you want humans to still exist, be replaced, cyborgized and/or uploaded/simulated?
  3. Do you want humans or machines in control?
  4. Do you want AIs to be conscious or not?
  5. Do you want to maximize positive experiences, minimize suffering or leave this to sort itself out?
  6. Do you want life spreading into the cosmos?
  7. Do you want a civilization striving toward a greater purpose that you sympathize with, or are you OK with future life forms that appear content?

Depending on your answers this might lead to one of the following scenario



A verbal description of the scenarios is given below, type the name of the scenario into the left field, if you want more scenarios to be shown increase the “Show entries” entry


Data religion: Dataism

A data based religion called Dataism is a concept described by Harari in Homo Deus: A brief history of tomorrow (Harari 2016) and says:


In data we trust


Dataism: only wild fantasy?


Economists interpret economy as data processing system

Career: Oxford seeks AI ethics professor

Binet, Alfred, and Th Simon. 1916. “New Methods for the Diagnosis of the Intellectual Level of Subnormals.(L’Année Psych., 1905, Pp. 191-244).” Williams & Wilkins Co.

Goodfellow, Ian J, Jonathon Shlens, and Christian Szegedy. 2014. “Explaining and Harnessing Adversarial Examples.” arXiv Preprint arXiv:1412.6572.

Gottfredson, Linda S. 1997. “Mainstream Science on Intelligence: An Editorial with 52 Signatories, History, and Bibliography.” Citeseer.

Harari, Yuval Noah. 2016. Homo Deus: A Brief History of Tomorrow. Random House.

Kahneman, Daniel. 2011. Thinking, Fast and Slow. Macmillan.

Russakovsky, Olga, Jia Deng, Hao Su, Jonathan Krause, Sanjeev Satheesh, Sean Ma, Zhiheng Huang, et al. 2015. “ImageNet Large Scale Visual Recognition Challenge.” International Journal of Computer Vision (IJCV) 115 (3): 211–52. https://doi.org/10.1007/s11263-015-0816-y.

Tegmark, Max. 2017. Life 3.0: Being Human in the Age of Artificial Intelligence. Knopf.


  1. American pioneer in the field of computer gaming and artificial intelligence https://en.wikipedia.org/wiki/Arthur_Samuel

  2. professor of Computer Science and Machine Learning at Carnegie Mellon http://www.cs.cmu.edu/~tom/

  3. 17*35=595

  4. Knowledge neglect https://en.wikipedia.org/wiki/Knowledge_neglect

  5. Die drei letzten Bundespräsidenten: Steinmeier, Gauck, Wulf

  6. https://www.anesi.com/bayes.htm

  7. Pingo question: What is the probability that the cab involved in the accident was Blue rather than Green? https://pingo.coactum.de/questions

  8. Screening mammography aims to identify breast cancer at earlier stages of the disease, when treatment can be more successful. Despite the existence of screening programmes worldwide, the interpretation of mammograms is affected by high rates of false positives and false negatives. Here we present an artificial intelligence (AI) system that is capable of surpassing human experts in breast cancer prediction. To assess its performance in the clinical setting, we curated a large representative dataset from the UK and a large enriched dataset from the USA. We show an absolute reduction of 5.7% and 1.2% (USA and UK) in false positives and 9.4% and 2.7% in false negatives. We provide evidence of the ability of the system to generalize from the UK to the USA. In an independent study of six radiologists, the AI system outperformed all of the human readers: the area under the receiver operating characteristic curve (AUC-ROC) for the AI system was greater than the AUC-ROC for the average radiologist by an absolute margin of 11.5%. We ran a simulation in which the AI system participated in the double-reading process that is used in the UK, and found that the AI system maintained non-inferior performance and reduced the workload of the second reader by 88%. This robust assessment of the AI system paves the way for clinical trials to improve the accuracy and efficiency of breast cancer screening. https://www.nature.com/articles/s41586-019-1799-6

  9. Abstract With an estimated 160,000 deaths in 2018, lung cancer is the most common cause of cancer death in the United States1. Lung cancer screening using low-dose computed tomography has been shown to reduce mortality by 20–43% and is now included in US screening guidelines. Existing challenges include inter-grader variability and high false-positive and false-negative rates. We propose a deep learning algorithm that uses a patient’s current and prior computed tomography volumes to predict the risk of lung cancer. Our model achieves a state-of-the-art performance (94.4% area under the curve) on 6,716 National Lung Cancer Screening Trial cases, and performs similarly on an independent clinical validation set of 1,139 cases. We conducted two reader studies. When prior computed tomography imaging was not available, our model outperformed all six radiologists with absolute reductions of 11% in false positives and 5% in false negatives. Where prior computed tomography imaging was available, the model performance was on-par with the same radiologists. This creates an opportunity to optimize the screening process via computer assistance and automation. While the vast majority of patients remain unscreened, we show the potential for deep learning models to increase the accuracy, consistency and adoption of lung cancer screening worldwide. Website: https://www.nature.com/articles/s41591-019-0447-x

  10. Homepage of ILSVRC http://www.image-net.org/challenges/LSVRC/

  11. Homepage of ImageNet http://www.image-net.org

  12. https://deepmind.com/blog/article/alphago-zero-starting-scratch

  13. Homepage of DeepMind https://deepmind.com

  14. Pingo question: Ihre Position zur AGI? https://pingo.coactum.de/questions

  15. “Dataism was born from the explosive confluence of two scientific tidal waves. In the 150 years since Charles Darwin published On the Origin of Species, the life sciences have come to see organisms as biochemical algorithms. Simultaneously, in the eight decades since Alan Turing formulated the idea of a Turing Machine, computer scientists have learned to engineer increasingly sophisticated electronic algorithms. Dataism puts the two together, pointing out that exactly the same mathematical laws apply to both biochemical and electronic algorithms. Dataism thereby collapses the barrier between animals and machines, and expects electronic algorithms to eventually decipher and outperform biochemical algorithms.” Harari, Yuval Noah. Homo Deus: A Brief History of Tomorrow . HarperCollins. Kindle Edition.

Reuse

Text and figures are licensed under Creative Commons Attribution CC BY 4.0. The figures that have been reused from other sources don't fall under this license and can be recognized by a note in their caption: "Figure from ...".

Citation

For attribution, please cite this work as

Sterr (2020, Jan. 21). Uwe's Blog: Machine learning, shall we?. Retrieved from http://uwesterr.de/posts/2020-01-21-machine-learning-shall-we/

BibTeX citation

@misc{sterr2020machine,
  author = {Sterr, Uwe},
  title = {Uwe's Blog: Machine learning, shall we?},
  url = {http://uwesterr.de/posts/2020-01-21-machine-learning-shall-we/},
  year = {2020}
}