Posts

LNAI 9605 Machine Learning for Health Informatics available

14.12.2016 LNAI 9605 just appeared

Machine Learning for Health Informatics Lecture Notes in Artificial Intelligence LNAI 9605

Holzinger, Andreas (ed.) 2016. Machine Learning for Health Informatics: State-of-the-Art and Future Challenges. Cham: Springer International Publishing, doi:10.1007/978-3-319-50478-0

[book homepage]

Machine learning (ML) is the fastest growing field in computer science, and Health Informatics (HI) is amongst the greatest application challenges, providing future benefits in improved medical diagnoses, disease analyses, and pharmaceutical development. However, successful ML for HI needs a concerted effort, fostering integrative research between experts ranging from diverse disciplines from data science to visualization.

Tackling complex challenges needs both disciplinary excellence and cross-disciplinary networking without any boundaries. Following the HCI-KDD approach, in combining the best of two worlds, it is aimed to support human intelligence with machine intelligence.

This state-of-the-art survey is an output of the international HCI-KDD expert network and features 22 carefully selected and peer-reviewed chapters on hot topics in machine learning for health informatics; they discuss open problems and future challenges in order to stimulate further research and international progress in this field.

NIPS 2016 is over

A crazy 5700-people event is over: NIPS 2016 in Barcelona. Registration on Sunday, 4th December, on Monday, 5th traditionally the tutorials were presented concluded by the first keynote talk given by Yann LeCun (now director at Facebook AI research) and completed by the official opening and the first poster presentation.  On Tuesday, Dec 6th, after starting with a keynote by Drew Purves (Google Deep Mind), parallel tracks on clustering and graphical models took place concluded by a keynote given by Saket Nevlakha (The Salk Institute) and completed by parallel tracks on deep learning and machine learning theory and poster sessions and demonstrations. Wednesday was openend by a keynote from Kyle Cranmer (New York University), the award talk “matrix completion has no spurious local min” and dominated by parallel tracks on algorithms and applications, concluded by a keynote by Marc Raibert (Boston Dynamics) who presented advances in latest robotic learning, and parallel tracks on deep learning and optimization, completed by the poster sessions with cool demonstrations. The Thursday was opened by a keynote fromm Irina Rish (IBM) and Susan Holmes (Stanford), followed by parallel tracks on interpretable models and cognitive neuroscience, concluded by various symposia until 21:30! Friday and Saturday were the whole day workshops – the sunday was reserverd for recreation on the sand beach of Barcelona 🙂

NIPS is definitely the most exciting conference with amazing variety on topics and themes revolving in machine learning with all sorts of theory and applications.

nips-2016-barcelona-machine-learningnips-2016-barcelona-machnine-learning-gamification

Machine Learning with Fun

Google Research hosts a number of very interesting so-called A.I. experiments. There you can play with machine learning algorithms in a very amusing way. A recent example is QUICK, DRAW *). This is an online guessing game that challenges humans to hand sketch (called doodles) a given object. The game uses a  neural network to learn from the input data

https://quickdraw.withgoogle.com

which is part of the A.I. Experiments platform:

https://aiexperiments.withgoogle.com

and here the explanatory video:
https://www.youtube.com/watch?v=oOwfiYnRi5c

Have fun and enjoy!

Here you see more than 100.000 hedgehog drawings made by humans on the internet:

https://quickdraw.withgoogle.com/data/hedgehog

*) not to be confused with QuickDraw [1], which is a sketch-based drawing tool facilitating to draw precise geometry diagrams,  and can automatically recognize sketched diagrams containing components such as line segments and circles, infer geometric constraints relating recognized components, and use this information to “beautify” the sketched diagram. This “Beautification” is based on an algorithm that iteratively computes various sub-components of the components using an extensible set of deductive rules.

[1] Cheema, S., Gulwani, S. & Laviola, J. QuickDraw: improving drawing experience for geometric diagrams. Proceedings of the SIGCHI Conference on Human Factors in Computing Systems, 2012. ACM, 1037-1064. doi: 10.1145/2207676.2208550

[2] https://experiments.withgoogle.com/ai

Obama on humans-in-the-loop

How artificial intelligence will affect jobs

In an discussion with Barack OBAMA [1] on how artificial intelligence will affect jobs, he emphasized how important human-in-the-loop machine learning will become in the future. Trust, transparency and explainabiltity will be THE driving factors of future AI solutions! The discussion interview was led by the Wired [2] Editor Scott DADICH, and MIT Media Lab [3] Director Joi ITO. I recommend my students to watch the full video. Barack Obama demonstrates a  good understanding of the field and indicates indirectly the importance of our research in the the human-in-the-loop approach [4], despite all progress towards fully automatic approaches and autonomous systems.

More information see:

[1] Barack Obama was the 44th President of the United States of America and was in office from January, 20, 2009 to January, 20, 2017. He was born August, 4, 1961 in Honolulu (Hawaii)

[2] Wired is a monthly tech magazine which reports since 1993 on how emerging technologies may affect culture, politics, economics. Very interesting to note is that Wired is known for coning the popular terms “long tail” and “crowdsourcing”. https://www.wired.com

[3] The MIT Media Lab is an interdisciplinary research lab at the Massachusetts Institute of Technology in Cambridge (MA), which is part of the Boston metropolitan area in the north, just across the Charles River – not far way from the Harvard Campus.

[4] Holzinger, A., Plass, M., Holzinger, K., Crisan, G.C., Pintea, C.-M. & Palade, V. 2017. A glass-box interactive machine learning approach for solving NP-hard problems with the human-in-the-loop. arXiv:1708.01104

 

 

Portfolio Items