The Winner of SAIS best AI Master’s Thesis Award 2014

We are happy to announce that the winner of the Swedish AI Society’s prize for best AI Master’s Thesis 2014 is Daniel de Leng, for his thesis work conducted at Linköping University. The title of the thesis is “Extending semantic matching in DyKnow to handle indirectly-available streams”. The supervisor is Fredrik Heintz, Linköping University and the examiner is Professor John-Jules Meyer, Utrecht University.

The motivation is: The award goes to Daniel de Leng for a thesis, which presents extensive work that demonstrates skills in both software development for robotics, formal theory, and formal languages, all useful building blocks for artificial intelligence. A strength, and which proves the soundness of his results, is that it goes all the way from theory to implementation and evaluation. The results fit nicely into a use context, and advance both theoretical work on semantic information integration for stream reasoning by taking into consideration indirectly-available streams, and practical implementation as the first integrated version of DyKnow-ROS, which can be used for empirical experimentation. In this context, it is clear that the results open up several avenues of new paths for research. The text is well written, showing results with high level of maturity and has resulted in a scientific publication, which further shows that it is of high quality.

The submitted theses were read and evaluated by a group of expert reviewers. A combination of four criteria was used to determine the winner: maturity of content, scientific value, writing style, and relevance of the content to the AI area. All submitted theses were of very high quality but the chosen thesis stood out when evaluating all criteria.

The thesis will be presented at the SAIS workshop, which is taking place the coming two days (May 22-23) at KTH, Stockholm. For more information and schedule, please, visit

http://www.sais.se/sais2014/

Helena Lindgren on behalf of the SAIS board

Posted in exjobb, sais | Comments Off

The SAIS AI Master’s Thesis Award 2014 – Call for nominations

In order to stimulate work in the area of artificial intelligence at the undergraduate level of education, SAIS has created a special yearly award for the best Master’s thesis (examensarbete) in the area of artificial intelligence. The winner will be awarded travel funds of 5000SEK, and is in addition expected to attend the SAIS Workshop to present the winning contribution (travel, hotel and registration paid by SAIS).

To nominate a Master’s Thesis for the award, the following rules apply:

  • The thesis is in one of the main areas of artificial intelligence research.
  • The thesis has been presented officially during the period of  January 1, 2013 – January 31, 2014.
  • An electronic copy of the thesis (PDF) and a short written motivation by the official examiner or primary supervisor should be sent to mthprize@sais.se, include the terms “SAIS” and “Master’s Thesis award” in the subject line, no later than February 28, 2014.  The nomination should include a short motivation why the thesis should be selected as the winner and state its contributions to AI.
  • The nominating person must be a member of SAIS.
  • Only one thesis nomination per nominating person is permitted.

The SAIS board will review the nominated Master’s theses and determine a winner based on factors such as novelty, thoroughness, readability, scientific value and relevance to artificial intelligence both with respect to research and to potential applications. The decision by the SAIS board will be final. The selected winner will be announced in time for the SAIS Workshop, where the winner will be expected to present their work.

Fredrik Heintz – President of SAIS
Eva Blomqvist – Secretary of SAIS

Posted in exjobb, sais | Comments Off

SAIS exjobbspris 2013 – vinnaren utsedd

Vinnare av SAIS exjobbspris 2013 är:
Andrea Baisero, KTH
för examensarbetet med titeln “Encoding Sequential Structures using Kernels - Path-Based Feature Space Embeddings for Sequential Data.”
(Nominerad av: Danica Kragic, KTH)

Motiveringen till priset lyder:
“The award goes to Andrea Baisero for his thesis entitled “Encoding
Sequential Structures using Kernels – Path-Based Feature Space
Embeddings for Sequential Data”. The thesis is of high theoretical
and mathematical strength, and it is presented in a report that is
thorough and well-written. Baisero has developed a new type of feature
space for sequential data implicitly defined through an inner-product.
This is an important area of theoretical research, related to Artificial
Intelligence and in particular Machine Learning, where results are
applicable in many application domains where data naturally is presented
in terms of sequential structures, such as bioinformatics, Internet
security and language processing. The results are novel and presents a
significant contribution towards dealing with sequential data.”

Det vinnande exjobbet finns tillgängligt här. Vinnaren kommer att presentera sitt arbete vid SCAI 2013.

Totalt fick SAIS styrelse in 5 nomineringar till årets pris. Varje exjobb utvärderades av tre oberoende granskare, varefter styrelsen utsåg en vinnare baserat på granskningsresultaten. De nominerade var

  • Shameem Ahamed Puthiya Parambath, Umeå universitet – “Topic Extraction and Bundling of Related Scientific Articles” (Nominerad av: Henrik Björklund)
  • Andrea Baisero, KTH - “Encoding Sequential Structures using Kernels - Path-Based Feature Space Embeddings for Sequential Data.” (Nominerad av: Danica Kragic)
  • Ayne Beyenes och Tewelle Welemariams, BTH – “Concept Drift in Surgery Prediction” (Nominerad av: Niklas Lavesson och Marie Persson)
  • Karl Jansson, Stockholms universitet – “Performance study of using the Direct Compute API for implementing Support vector machines on GPUs” (Nominerad av: Henrik Boström)
  • Anders Westling, KTH – “Sentiment Analysis of Microblog Posts from a Crisis Event using Machine Learning” (Nominerad av: Joel Brynielsson)

Eva Blomqvist
Sekreterare, SAIS

Posted in exjobb, sais, SAIS exjobbspris, sverige | Tagged | Comments Off

Renässans för Artificiell Intelligens

Området Artificiell Intelligens är sedan ett antal år föremål för ett kraftigt ökande intresse.

Resultaten från de första snart sextio åren av forskningsfältet artificiell intelligens har fått ett blandat bemötande. Å ena sidan menar många kritiker att trots de enorma summor forskningsmedel som lagts ner inom området under denna tid, så har inte de storslagna löftena om datorer som beter sig intelligent infriats.

Å andra sidan splittrades fältet relativt omgående upp i ett antal subdiscipliner: kunskapsrepresentation, maskininlärning, planering, datorseende, taligenkänning, naturligt språk, robotik, och många fler. Inom varje område har stora framsteg gjorts, och datorer har stadigt lyckats klara av fler och fler saker som tidigare krävt mänsklig inblandning. De utvecklade teknikerna har smugit sig på oss i vardagen utan att vi knappt märkt det: telefoner och telefonsvarare som man kan prata med, program som känner igen ansikten, gps:er som kan rekommendera snabbaste väg, gräsklippare och dammsugare som sköter sig själva, och hjälp med översättning mellan olika språk. I industrin finns ännu fler exempel på metoder som hjälper till att planera, optimera, reglera, och övervaka. De flesta av dessa hjälpmedel som översvämmar vår vardag stammar på något sätt från forskningen inom artificiell intelligens. Men det finns ett problem: Ingen av metoderna uppfattas som “intelligent”. De bygger “bara” på matrismultiplikation, statistik, mer eller mindre dum genomsökning av massa alternativ, diverse ingenjörsmässiga trick och knep, slimmad design, nya material, snabbare datorer, större minnen, och allt mer ihopkopplade system. Metoderna må vara användbara, men de uppfattas inte som intelligenta på samma sätt som en människa.

Så trots denna sannerligen enorma rad av succehistorier, så hade till för några år sedan hade begreppet artificiell intelligens en ganska negativ klang. Den intelligens man hade utlovat att på några få månader kunna härma med datorer, lyste fortfarande med sin frånvaro. Ännu hade ingen dator på allvar kunnat konkurera med verklig mänsklig kompetens. Men 2005 hände plötsligt något: Ett antal helt autonoma robotbilar hade lyckats ta sig genom ett ökenlandskap i USA (http://en.wikipedia.org/wiki/DARPA_Grand_Challenge). 2007 upprepades bedriften i (en för ändamålet uppbyggd) stadsmiljö där bilarna var tvungna att följa trafikregler. Sedan 2012 tillåter flera delstater i USA att autonoma bilar kör i den
vanliga trafiken (http://en.wikipedia.org/wiki/Google_driverless_car). Och 2011 hände något igen: Datorsystemet Watson lyckades vinna i finalen över människor i Jeopardy (http://www-03.ibm.com/innovation/us/watson). Plötsligt har opinionen vänt, och begreppet artificiell intelligens har fått återupprättelse. Men hur gick det till, vad är det för banbrytande upptäckt som plötsligt gjort dessa bedrifter möjliga?

Hemligheten är inte en enda upptäckt. De nämnda exemplen är hybridsystem som kombinerar tekniker från flera olika discipliner. Robotbilarna kräver reglerteknik, mekanik, datorseende, lokalisering, och planering. Watson innehåller komponenter från kunskapsrepresentation, ontologier, taligenkänning, naturligt språk-hantering, söktekniker, och maskininlärning. Kanske är det så att mänsklig “intelligens” inte heller är en enda förmåga, utan samverkan mellan flera olika. Var och en av de nämnda AI-områdena har kämpat med sina små delproblem, var och en inte uppfattad som särskilt intelligent. Först nu börjar vart och ett av områdena att bli tillräckligt mogna. Och när man nu sätter ihop dem till en samverkande helhet kan man få ett resultat som till och med kan bräcka en människa i vad som uppfattas som “intelligenta” förmågor. Vi har säkert bara sett början på detta – fler exempel kommer att följa i en accelererande takt.

Hela denna historia är samtidigt ett slående exempel på nyttan av grundforskning. Stora delar av AI-området har finansierats med långsiktiga medel utan krav på omedelbar tillämpbarhet. Efter ett antal “hypes” med högt uppskruvade förväntningar började finansiärerna dock tröttna, och många misströstade om att AI-forskningen skulle kunna leda till något användbart. Men när man höjer blicken så har användbara saker sipprat ut under lång tid, för att nu genomsyra hela samhället, om än kanske inte på det sätt som ursprungligen förväntats. Grundforskning är nödvändig om man vill komma framåt, och kan leda till verkligt genomgripande förändringar i samhället, men man måste ha tålamod innan man kan se resultaten.

Posted in Uncategorized | Comments Off

The SAIS AI Master’s Thesis Award 2013 – Call for nominations

In order to stimulate work in the area of artificial intelligence at
the undergraduate level of education, SAIS has created a special
yearly award for the best Master’s thesis (examensarbete) in the area
of artificial intelligence. The winner will be awarded travel funds of
5000SEK, and is in addition expected to attend the SCAI 2013 conference
to present the winning contribution (travel, hotel and registration paid by SAIS).

To nominate a Master’s Thesis for the award, the following rules apply:

  • The thesis is in one of the main areas of artificial intelligence research.
  • The thesis has been presented officially during the period of  January 1, 2012 – January 31, 2013.
  • An electronic copy of the thesis (PDF or Postscript) and a short written motivation by the official examiner or primary supervisor should be sent to the secretary of SAIS, Eva Blomqvist (evabl444<snabela> gmail.com - include the terms “SAIS” and “Master’s Thesis award” in the subject line), no later than February 28, 2013.  The nomination should include a short motivation why the thesis should be selected as the winner and state its contributions to AI.
  • The nominating person must be a member of SAIS.
  • Only one thesis nomination per nominating person is permitted.

The SAIS board will review the nominated Master’s theses and determine
a winner based on factors such as novelty, thoroughness, readability,
scientific value and relevance to artificial intelligence both with
respect to research and to potential applications. The decision by the
SAIS board will be final. The selected winner will be announced latest in early
June, and the prize will be awarded during the SCAI conference in Aalborg,
November 20-22 2013, where the winner will be expected to give a short
presentation about their work.

Fredrik Heintz – President of SAIS
Eva Blomqvist – Secretary of SAIS

Posted in exjobb, sais, SAIS exjobbspris, sverige | Comments Off

Patrick Doherty New President of ECCAI

It is my great pleasure to inform you that Patrick Doherty, Linköping University, is the new president of ECCAI, the European Coordinating Committee for Artificial Intelligence.

Congratulations Patrick!

Fredrik Heintz President of SAIS

Posted in Europe, sais | Comments Off

Report from ECCAIs General Assembly August 29, 2012

ECCAI, the European Coordination Committee for Artificial Intelligence, held its yearly General Assembly yesterday at ECAI in Montpellier. ECCAI consists of 29 member societies, including SAIS. Two of its most important activities are arranging ECAI and publishing AI Communications.

Highlights of the meeting are:

  • Patrick Doherty, Linköping University, is the new President of ECCAI. The General Assembly elects the board and the board then elects one of its members as President.
  • ECAI 2016 will be held in The Hague. They had strong competition from Moscow and Trondheim.
  • ECCAI and IJCAI has agreed to organize a joint IJCAI/ECAI conference every fourth year starting in 2018.

ECAI 2014 will be in Prague.

Posted in Europe, sais | Comments Off

Erik Sandewall receives ECCAI Distinguished Service Award

Erik Sandewall, Linköping University, receives the inaugural ECCAI Distinguished Service Award for his extraordinary contributions and outstanding service to the Artificial Intelligence community. The award was handed out yesterday at ECAI in Montpellier.

“Erik Sandewall can truly be considered as a pioneer and ambassador of Artificial Intelligence in both Europe and internationally with both a research and distinguished service record well known to several generations of researchers in AI. In Erik, we see a unique example of a cutting edge scientist and out-of-the-box thinker that has not only contributed to the area of Artificial Intelligence through many seminal research contributions, but also spent tremendous amounts of time promoting AI in Europe and elsewhere and contributing his ideas and competences generously to many other research groups in Europe in an effort to promote the science and success of AI. This has been done in a professional and selfless manner and in our opinion, we believe he is an ideal candidate for receiving the ECCAI Distinguished Service award.”
From the nomination by ECCAI Fellow Patrick Doherty and SAIS.

Posted in Europe, sais | Comments Off

ECAI 2012, 27-31 Augusti, Montpellier, France

Sista veckan i augusti samlas mer än 750 forskare, studenter och utövare inom AI i Montpellier, Frankrike, för den 20:e Europeiska AI-konferensen (ECAI). Av 563 bidrag (492 långa och 71 korta) blev 140 långa (28%) och 23 (32%) korta artiklar accepterade. Kunskapsrepresentation, maskininlärning och multiagentsystem är de områden med flest bidrag, precis som det har varit de senaste gångerna. Inbjudna talare är Wolfram Burgard, Probabilistic Techniques for Mobile Robot Navigation, Adnan Darwiche Generalized Decision Diagrams: The Game is not over yet!, Tom Mitchell Never Ending Learning och Michael Wooldridge Bad Equilibria, and What to do About Them.

Svenska bidrag till konferensen är

  • Macros, Reactive Plans and Compact Representations
    Christer Bäckström, Anders Jonsson and Peter Jonsson
  • From Macro Plans to Automata Plans
    Christer Bäckström, Anders Jonsson and Peter Jonsson
  • An O(n log n) Bound Consistency Algorithm for the Conjunction of an alldifferent and an Inequality between a Sum of Variables and a Constant, and its Generalization
    Nicolas Beldiceanu and Mats Carlsson and Thierry Petit and Jean-Charles Régin
  • Erik Sandewall är både nominerad till det nyinstiftade ECCAI Distinguished Service Award och inbjuden att hålla ett föredrag om A Perspective on the Early History of AI in Europe som en del av Turing and Anniversary Session.
  • Patrick Doherty och Fredrik Heintz ansvarar för ECAI Systems Demonstration Track.

ECAIs hemsida
Lista med accepterade artiklar
Beställ proceedings

Posted in conference, conference report, Europe | Comments Off

Report on the Dagstuhl seminar on AI and CI in games

Back in May, some of my colleagues and I arranged a seminar at the Dagstuhl castle in Germany on “Artificial and Computational Intelligence in Games”. It was quite an honour to be invited to organise a meeting at such a prestigious venue, especially as there aren’t that many of us working on AI in games. Yet. In any case, the meeting went very well, and really helped the 36 attendants to reach something like a joint vision about what the field is all about and where we are going. We organised the meeting into a number of workgroups, each including a handful of people discussing the state of the art and future research directions for a particular topic, such as pathfinding, player modelling or content generation.

The first outcome of the meeting is now available. It’s a “Dagstuhl Report”, or essentially a collection of abstracts:
http://drops.dagstuhl.de/opus/volltexte/2012/3651/
We are working on a longer edited volume, where each workgroup will write a 10-15 page chapter. However, that will not come out for a few months. In the meantime, if you are interested in what goes on in AI applied to games, feel free to have a look at the seminar report!

Posted in conference report | Comments Off