If you’ve been following this blog you’ll know that there have been great advances in the past few years with artificial image generation, to the stage where having a picture of something does not necessarily mean that it is real. Image advances are easy to talk about, as there’s something tangible to show, but there have been similar large leaps forward in other areas, particularly in voice synthesis and handwriting.
Earlier this month, Dove launched their new baby range with another of their fantastic adverts challenging stereotypes and questioning is there a “perfect mum”. As a mum myself I can relate to the many hilarious bloggers1 who are refreshingly honest about the unbrushed hair, lack of make-up, generally being covered in whatever substances your new tiny human decides to produce, and all other parenting frustrations. I’m really pleased that there are lots of women2 out there challenging the myths presented in the media – we don’t all have a team to make us beautiful, nor someone photo-shopping the results to perfection, and the pressure can be immense. This is where Dove’s campaign is fantastic. Rather than just creating a photoshoot with a model and doctoring the results, the image is actually completely artificial, having been generated by AI. Continue reading Artifical image creation takes another step forward in advertising
I love the fact that here in the UK everyone can be involved in shaping the future of our country, even if a large number of individuals choose not to and, in my eyes, if you don’t get involved then you don’t have the right to complain. While this is most generally applied to the election of our representatives from local parish councils to our regional MPs (or actually standing yourself)1 there are also a lot of other ways to be involved. In addition to raising issues with your local representative, parliament has cross bench committees that seek input from the public and to help create policy or consider draft legislation.
Our elected parliament is not made up of individuals who are experts in all fields. Even government departments are not necessarily headed by individuals with large amounts of relevant experience. It is critical that these individuals are informed by those with the experience and expertise in the issues that are being considered. Without this critical input, our democracy is weakened. Continue reading Submitting evidence to parliament committees
If you’re starting out in deep learning and would prefer a laptop over a desktop, basic research will lead you to a whole host of blogs, Q&A sites and opinions that basically amount to “don’t do it” and to get a desktop or remote into a server instead. However, if you want a laptop, whether this is for college, conferences or even because you have a job where you can work from anywhere, then there are plenty of options available to you. Here I’ll lay out what I chose and why, along with how it’s performing. Continue reading Choosing a Laptop for Deep Learning Development
The Science and Technology Select Committee here in the UK have launched an inquiry into the use of algorithms in public and business decision making and are asking for written evidence on a number of topics. One of these topics is best-practise in algorithmic decision making and one of the specific points they highlight is whether this can be done in a ‘transparent’ or ‘accountable’ way1. If there was such transparency then the decisions made could be understood and challenged.
It’s an interesting idea. On the surface, it seems reasonable that we should understand the decisions to verify and trust the algorithms, but the practicality of this is where the problem lies. Continue reading Algorithmic transparency – is it even possible?
Learning to play games has been a great test for AI. Being able to generalise from relatively simple rules to find optimal solutions shows a form of intelligence that we humans always hoped would be impossible. Back in 1997, when IBMs Deep Blue beat Gary Kasparov in chess1 we saw that machines were capable of more than brute force solutions to problems. 20 years later2 and not only has AI mastered Go with Google’s DeepMind winning 4-1 against the world’s best player and IBM’s Watson has mastered Jeopardy, there have also been some great examples of game play with many of the games I grew up playing: Tetris, PacMan3, Space Invaders and other Atari games. I am yet to see any AI complete Repton 2. Continue reading Anything you can do AI can do better (?): Playing games at a new level
“Pics or it didn’t happen” – it’s a common request when telling a tale that might be considered exaggerated. Usually, supplying a picture or video of the event is enough to convince your audience that you’re telling the truth. However, we’ve been living in an age of Photoshop for a while and it has (or really should!!!) become habit to check Snopes and other sites before believing even simple images1 – they even have a tag for debunked images due to photoshopping. Continue reading Artificial images: seeing is no longer believing
In September 2016, the ReWork team organised another deep learning conference in London. This is the third of their conferences I have attended and each time they continue to be a fantastic cross section of academia, enterprise research and start-ups. As usual, I took a large amount of notes on both days and I’ll be putting these up as separate posts, this one covers the morning of day 1. For reference, the notes from previous events can be found here: Boston 2015, Boston 2016.
When I attended the ReWork Deep Learning conference in Boston in May 2016, one of the most interesting talks was about the Echo and the Alexa personal assistant from Amazon. As someone whose day job is AI, it seemed only right that I surround myself by as much as possible from other companies. This week, after it being on back order for a while, it finally arrived. At £50, the Echo Dot is a reasonable price, with the only negative I was aware of before ordering being that the sound quality “wasn’t great” from a reviewer. Continue reading Amazon Echo Dot (second generation): Review
As part of a few hours catching up on machine learning conference videos, I found this great talk on what can go wrong with machine recommendations and how testing can be improved. Evan gives some great examples of where the algorithms can give unintended outputs. In some cases this is an emergent property of correlations in the data, and in others, it’s down to missing examples in the test set. While the talk is entertaining and shows some very important examples, it made me realise something that has been missing. The machine learning community, having grown out of academia, does not have the same rigour of developmental processes as standard software development.
Regardless of the type of machine learning employed, testing and quantification of results is all down to the test set that is used. Accuracy against the test set, simpler architectures, fewer training examples are the focal points. If the test set is lacking, this is not uncovered as part of the process. Models that show high precision and recall can often fail when “real” data is passed through them. Sometimes in spectacular ways as outlined in Evan’s talk: adjusting pricing for Mac users, Amazon recommending inappropriate products or Facebook’s misclassification of people. These problems are either solved with manual hacks after the algorithms have run or by adding specific issues to the test set. While there are businesses that take the same approach with their software, they are thankfully few and far between and most companies now use some form of continuous integration, automated testing and then rigorous manual testing.
The only part of this process that will truly solve the problem is the addition of rigorous manual testing by professional testers. Good testers are very hard to find, in some respect harder than it is to find good developers. Testing is often seen as a second class profession to development and I feel this is really unfair. Good testers can understand the application they are testing on multiple levels, create the automated functional tests and make sure that everything you expect to work, works. But they also know how to stress an application – push it well beyond what it was designed to do, just to see whether these cases will be handled. What assumptions were made that can be challenged. A really good tester will see deficiencies in test sets and think “what happens if…”, they’ll sneak the bizarre examples in for the challenge.
One of the most difficult things about being a tester in the machine learning space is that in order to understand all the ways in which things can go wrong, you do need some appreciation of how the underlying system works, rather than a complete black box. Knowing that most vision networks look for edges would prompt a good tester to throw in random patterns, from animal prints to static noise. A good tester would look of examples not covered by the test set and make sure the negatives far outweigh the samples the developers used to create the model.
So where are all the specialist testers for machine learning? I think the industry really needs them before we have (any more) decision engines in our lives that have hidden issues waiting to emerge…