This week, we’re looking at the Google Brain, which Microsoft says is outperformed by their own Adam’, why embracing hardware data can bring productions costs down, and the new ‘social’ robot Jibo and its emotional skill set. There’s some great reads on moral competence for social robots and the challenges of engineering ethics, and the fanciest personal data interface you’ve ever seen.
This Week in Context
Your Weekly Update on All Things Context, July 18 2014
We’re happy to announce we’ve moved to ‘arguslabs.com‘. So starting today, you’ll find the #WIC delivered into your mail box from @sentiance.com, if you haven’t yet, subscribe here. The weekly web edition you can now find online at www.arguslabs.com/blog. Happy reading!
Embracing Hardware Data
O’Reilly’s Radar looks at the collision of hardware and software through the eyes of a data scientist. Takeaway: ‘Design to disappear‘ becomes just as important for physical products, and not just from a user experience perspective. When you reduce the user interface on the device, you reduce manufacturing complexity – and thus manufacturing costs.
Engineering Ethics for Robots
In the ‘Moral Competence in Social Robots’, Matthias Scheutz argues that moral competence consists of four broad components: a moral core, as well as moral action, moral cognition and emotion, and moral communication. EE Times interviewed Scheutz regarding ethics for autonomous robots in healthcare and on the battlefield, and some of the challenges that robots’ designers and engineers face in terms of what decisions these robots will make.
“Pandora, for example, comes down to this thing that they call the music genome, which contains a summary of your musical tastes. To recommend a song, all you need is the last 10 songs you listened to — just to make sure you don’t keep recommending the same one again — and this music genome. You don’t need the list of all the songs you’ve been listening to.”
– Yves-Alexandre de Montjoye
quoted in Own Your Own Data by Larry Hardesty
Own Your Own Data: Sharing Code, Not Data
MIT researchers explain their openPDS (personal data store), a prototype system that stores data from digital devices in a single location – cloud or ‘self-hosted’ – specified by the user. One of the benefits of openPDS is it requires applications to specify what information they need and how it will be used. OpenPDS preserves all of the potentially useful data in a repository controlled by the end user, rather than by the application developer or service provider.
Wired has a look ‘inside the artificial brain that’s remaking the Google empire‘. Think deep learning, AIaS (artificial intelligence as a service) and how ‘Google Brain’ makes that the giant is no longer a search company, but a machine-learning company.
Battle of the Brains: Enter Adam
Microsoft is challenging Google’s artificial brain with its own deep-learning system nicknamed Adam, that is said to have better optimised the way its machines handle data and fine-tunes the communications between them. Microsoft says that this system – which puts less computing effort into avoiding collisions – could help its neural networks more quickly and more accurately train themselves.
A Personal Data Interface That Even The Doctor (EMH) Would Fancy
The Making of AprilZero is a well-explained look at the thinking and design process behind an incredible interactive visualisation of 110 days of personal data gathered from various smartphone apps. Admittedly, this personal data dashboard is astonishingly beautiful, but also a tad complex. The latter is of course a trade off you need to make, and Anand clearly decided he wanted – visually stunning – in depth insights.
Jibo, an Ambitious and Affordable Home Robot
Jibo, ‘the world’s first family robot’, will come packed with educational, entertainment, communications and photography applications. The 11-inch tall robot can act out stories, snap family images, flag appointments, play voicemails and read emails aloud, and acts as a telepresence robot – all for $499 for those who ‘pre-order’ on Indiegogo. (via Joren)
Will Jibo the Robot be Emotion-Aware?
Jibo will be able to express – convey – emotion by adjusting its ‘posture’ and movements, yet unlike the much more expensive Pepper, Jibo is not ’emotion-aware’. Jibo’s not able to detect our emotions or respond directly to them. Out of the box, that is. However, as the robot will come with 2 color stereo cameras and is tuned in to your data feeds already, who knows, by then he might be able to ‘see’ and ‘know’ how we feel? After all, that’s just a few firmware upgrades (and some hard AI work) away. Jibo Inc will also open up Jibo for external developers, who will be able to create and sell applications that give Jibo new capabilities. So maybe we should order a developer-Jibo for the Argus Labs office? 😉
Papers, Talks & Research
- EMOTEX: Detecting Emotions in Twitter Messages (emotion, paper)
- Fuzzy Model on Human Emotions Recognition (emotion, paper)
- Emotion Pattern Recognition Using Physiological Signals (smart sensors, paper)
- BB got your Beat: Visualising Heartbeats within a Biosensing Artwork (visualisation, paper)
- Exploring indicators from keyboard and mouse interactions to predict the user affective state (emotion, poster)
- Affective Computing: SMART Interactions of Machines with Humans (H2M interaction, paper)
- Music Genre and Emotion Recognition Using Gaussian Processes (music emotion classification, paper)
- Music Emotion Classification Based on Music Highlight Detection (music emotion classification, abstract)
- The need for moral competency in autonomous agent architectures (ethics, paper)
- Face detection without bells and whistles (face detection, paper)
- Consciousness is a Mathematical Pattern: Max Tegmark at TEDxCambridge 2014 (physics, TEDx video)
If you have some time to spare this weekend, do check out the Best of Strata Santa Clara 2014. ‘People are Data Too’ by Farrah Bostic and ‘The Art of Good Practise’ by Rodney Mullen are both must-see talks.
Enjoy the weekend, see you next week.