What happens when automated algorithms interact in ways unforeseen by their creators? One example is shown by two online book-pricing algorithms, automatically adjusting prices in response to each other’s price adjustments, in an infinitely-ascending dance of the algorithms:
A few weeks ago a postdoc in my lab logged on to Amazon to buy the lab an extra copy of Peter Lawrence’s The Making of a Fly – a classic work in developmental biology that we – and most other Drosophila developmental biologists – consult regularly. The book, published in 1992, is out of print. But Amazon listed 17 copies for sale: 15 used from $35.54, and 2 new from $1,730,045.91 (+$3.99 shipping).
Even rather simple adaptive systems can exhibit unintended global behaviours. Full story here. (HT: ER)
The Department of Informatics at King’s College London currently has a number of academic vacancies, in the Agents and Intelligent Systems (AIS) Group and in the Software Modeling and Applied Logic (SMAL) Group. The vacancies are for people at all academic levels – Lecturers (Assistant Professors), Senior Lecturers (Associate Professors), and full Professors.
Further details are available at these sites:
Lectureships in SMAL, and
The closing date is 20 May 2011.
A search algorithm is a computational procedure (an algorithm) for finding a particular object or objects in a larger collection of objects. Typically, these algorithms search for objects with desired properties whose identities are otherwise not yet known. Search algorithms (and search generally) has been an integral part of artificial intelligence and computer science this last half-century, since the first working AI program, designed to play checkers, was written in 1951-2 by Christopher Strachey. At each round, that program evaluated the alternative board positions that resulted from potential next moves, thereby searching for the “best” next move for that round.
The first search algorithm in modern times dates from 1895: a depth-first search algorithm to solve a maze, due to amateur French mathematician Gaston Tarry (1843-1913). Now, in a recent paper by logician Wilfrid Hodges, the date for the first search algorithm has been pushed back much further: to the third decade of the second millenium, the 1020s. Hodges translates and analyzes a logic text of Persian Islamic philosopher and mathematician, Ibn Sina (aka Avicenna, c. 980 – 1037) on methods for finding a proof of a syllogistic claim when some premises of the syllogism are missing. Representation of domain knowledge using formal logic and automated reasoning over these logical representations (ie, logic programming) has become a key way in which intelligence is inserted into modern machines; searching for proofs of claims (“potential theorems”) is how such intelligent machines determine what they know or can deduce. It is nice to think that theorem-proving is almost 1000 years old.
B. Jack Copeland : What is Artificial Intelligence?
Wilfrid Hodges : Ibn Sina on analysis: 1. Proof search. or: abstract state machines as a tool for history of logic. pp. 354-404, in: A. Blass, N. Dershowitz and W. Reisig (Editors): Fields of Logic and Computation. Lecture Notes in Computer Science, volume 6300. Berlin, Germany: Springer. A version of the paper is available from Hodges’ website, here.
Gaston Tarry : La problem des labyrinths. Nouvelles Annales de Mathématiques, 14: 187-190.
Welcome to InKings! This post is a quick shout to our work on automated trading systems and automated mechanism design.
A few weeks ago I gave an invited talk on competition between financial markets to a conference on Algorithmic Trading held in London. The conference was jointly organized by the Financial Computing Centre of University College London and LMAX, an innovative online financial trading marketplace. I spoke mainly about the CAT Market Design Tournament, an international research tournament established in 2007 to encourage research into the design of adaptive and automated market mechanisms. With the rapid recent growth of online financial trading, electronic marketplaces have proliferated, some under the labels “alternative trading platforms,” “upstairs markets,” or “dark pools of liquidity” (since many of these markets permit anonymous trades).
Competition between traditional stock exchanges and these alternative venues has intensified, and the need for adaptive responses by online markets to dynamic competitive conditions has therefore increased. The CAT Tournament has sought, through crowd-sourced research, to provide the theoretical and deployment underpinnings for the development of the science of automated and adaptive marketplace design. The tournament has been run successfully four times since 2007, under the aegis of the Trading Agent Competition, and has attracted entrants from every continent. The Tournament was undertaken as part of a major research project funded by the EPSRC, undertaken between the Universities of Birminghan, Liverpool and Southampton, with the involvement of Brooklyn College New York. A special issue of the academic journal, E-Commerce Research and Applications, devoted to reseach arising from the CAT Tournament, is due to appear in 2012.
The UCL/LMAX conference on Algorithmic Trading was very successful, and speakers pointed to the opportunities and risks arising from the deployment of sophisticated computing technology in economic and financial domains. Event co-sponsor LMAX is a newly-established multilateral online financial trading marketplace, created with a goal of enabling retail financial traders to undertake automated software trading through an open API. Videos of the talks can be found here (you will first have to register with LMAX). I recommend watching the talk by Dave Cliff of Bristol University, who summarizes compellingly the current lack of understanding we have of the workings of automated trading systems: we may be sitting at the confluence of four centuries of macro-economic crises and two centuries of technological disasters.
This blog provides an informal forum for discussion and debate surrounding issues related to Informatics at King’s College London. The postings on this site reflect individual views and do not necessarily represent the position of King’s College London.
Informatics at King’s covers Computer Science, Robotics, Telecommunications and Bioinformatics!