Deep Learning: John Kelleher **

This is an entry in a series from the MIT Press that selects a small part of a topic (in this case, a subset of artificial intelligence) and gives it an 'essential knowledge' introduction. The problem is, there seems to be no consistency over the target audience of the series.

I previously reviewed Virtual Reality in the same series and it kept things relatively simple and approachable to the general reader, even if it did overdo the hype. This book by John Kelleher starts gently, but by about half way through it has become a full-blown simplified textbook with far too much in-depth technical content. That's exactly what you don't want in a popular science title.

What we get is plenty of detail of what deep learning-based systems are and how they work at the technical level, but there is practically nothing on how they fit with applications (unless you count playing games), which are described but not really explained, nor is there anything much on the problems that arise when deep learning is used for real world applications. There is a passing reference, admittedly to the difficulties of understanding how a deep learning AI system came to a decision and how this clashes with the EU's GDPR requirement for transparency and explanation, but if feels more like this is done to criticise the naivety of the legislation than the danger of using such systems.

Similarly, I saw nothing about the dangers of deep learning systems using big data picking up on correlations that don't involve any causal link, nor does the book discuss the long tail problems that arise with inputs that are relatively uncommon and so are unlikely to turn up in the training data. Similarly we read nothing about the dangers of adversarial attacks, which can fool the systems into misinterpreting inputs with tiny changes, or the difficulties such systems have with real, messy environments as opposed to the rigid rules of a game.

Overall, the book is both pitched wrong and doesn't cover the aspects that really matter to the public. It may well do fine as an introductory text for a computer science student, but that doesn't fit with the blurb on the back, which implies it is for public consumption.

Paperback:

Kindle:

Using these links earns us commission at no cost to you

Review by Brian Clegg

Comments

Mathematics with Love – Mary Stopes-Roe *****

Admittedly it’s early days (this review is written in January), but this, for me, is the surprise hit of the year so far! I approached this book with trepidation, but found it absolutely delightful. It is described on the cover as the “courtship correspondence of Barnes Wallis, inventor of the bouncing bomb”, and contains a series of letters between Wallis and his cousin and eventual wife Molly Bloxham, along with some useful annotation by their daughter, Mary. The courtship itself is not without difficulties, as Wallis was 18 years older than the 17-year-old Molly at the start of the correspondence, and her father, not surprisingly, wasn’t too pleased about the interest of such an elderly suitor, but that isn’t the only reason the letters are interesting – it’s also because of maths, and Wallis’s position in the UK as the engineering hero of the Second World War. (Incidentally, it seemed very strange to see letters addressed to “Barnes” – I had always assumed Barnes Wallis was a ...

Data Empire - Roopika Risam ****

The central thesis presented by Roopika Risam is that information gives us (and particularly countries) the power to organise, control and dominate others. Although I have a couple of issues with the presentation, this is a genuinely interesting trip through the history of our use of stored information from the earliest tallies to the latest information technology. I loved a quote from Lisa Gitelman that data is is always 'cooked' so 'raw data is an oxymoron'. This neatly underlines Risam's thesis that data and information are not neutral facts, but rather tools that (like everything from fire to electronics) can be used for good or evil. As we are taken through the historical context, it can sometimes be a little difficult to judge whether Risam regards a particular example as bad or good, even when the outcome is disastrous. One thing I didn't like too much is the adherence to a popular science writing approach that has got distinctly hackneyed: opening chapte...

Andrew Jaffe - Five way interview

Andrew Jaffe is professor of astrophysics and cosmology at Imperial College, London and director of the Imperial Centre for Inference and Cosmology. His new book is The Random Universe. Why science? I’ve always been interested in science, in particular in astronomy, astrophysics, and space. One of my earliest memories - back in nursery school in New Jersey, I think - was watching one of the moon launches. I wanted that excitement to be part of my life! I never got to be an astronaut, but I did get to be part of the Planck Satellite team, and was privileged to be able to travel to the ESA Spaceport in French Guiana to watch the launch. In between, I was lucky enough to have a supportive family, get a good education, and find inspiring teachers, mentors, and collaborators. They helped me model the universe, and helped me learn how to refine those models in the face of experimental and observational evidence. That is, they taught me to be a scientist. Why this book? The Random...

Search This Blog