Authorea

Joe Corneli whitespace about 9 years ago

Commit id: e749f4543e04aca740c8cf06cadedb9d90f08e6f

deletions | additions

\begin{quote} {\em Identify a definition of serendipity that your system should satisfy to be considered serendipitous.}\end{quote} This is as As above. %% This situation can be pictured schematically as follows:

@inproceedings{schmidhuber2007simple, title={Simple algorithmic principles of discovery, subjective beauty, selective attention, curiosity \& creativity}, author={Schmidhuber, J{\"u}rgen}, booktitle={Discovery Science}, pages={26--38}, year={2007}, organization={Springer} } @book{bergson1983creative, title={Creative evolution}, author={Bergson, Henri},

discovery. Here, we revisit each of these criteria and briefly summarise how they can be thought about from a computational point of view, again focusing on examples. We then present a thought experiment\textbf{[Is that really what it is?]} that uses evaluates the ideas to develop described above in the course of developing a new system design. % \input{writers-workshop-background-long}

The features of our model match Merton's \citeyear{merton1948bearing} earlier description quite well: $T$ is the unexpected observation; $T^\star$ highlights its interesting or anomalous features and casts it as ``strategic data''; and the result $R$ may include updates to $p$ or $p^{\prime}$ that inform further phases of research. Connections The connection to the core key condition and components of serendipity introduced in our literature survey are as follows: % \textbf{Focus shift}. This The \textbf{focus shift} corresponds to the identification of $T^\star$, which is common to bothsides of the diagram. discovery and the invention phase. If the process operates in an ``online'' manner, $T^\star$ may bethought of as an evolving vector of interesting possibilities. % \textbf{Prepared mind}. This The \textbf{prepared mind} corresponds to the prior training $p$ and $p^{\prime}$ in our diagram. % \textbf{Serendipity trigger}. This corresponds to the stimulus The \textbf{serendipity trigger} is denoted by $T$ in our diagram. % \textbf{Bridge}. This corresponds to The \textbf{bridge} is comprised of the actions based on $p^{\prime}$ that are taken on $T^\star$ leading to the \textbf{result} $R$.% \textbf{Result}. This corresponds to our $R$. Note that $R$ may imply updates to $p$ or $p^{\prime}$ in further phases of research. In addition, Although they do not directly figure in our definition, the supportive dimensions and factors can be interpreted using this schematic, as follows: % \textbf{Chance}. One From the point of view of this model, $T$ is indeterminate. Furthermore, one must assume that relatively few of triggers $T^\star$ that are identified as interesting actually lead to useful results; in other words, the process is fallible. fallible and \textbf{chance} is likely to play a role. % \textbf{Curiosity}. The prior training $p$ causes interesting features to be extracted, even if they are not necessarily useful; $p^{\prime}$ asks how these features \emph{might} be useful. These routines suggest the relevance of a computational model of \textbf{curiousity}. One existing algorithmic approach is developed by \citeA{schmidhuber2007simple}. % \textbf{Sagacity}. Rather than a simple look-up rule, $p^{\prime}$ involves creating new knowledge. A simple example is found in clustering systems, which generate new categories on the fly. A more complicated example, necessary in the case of updating $p$ or $p$ is automatic programming. There is ample room for \textbf{sagacity} in this affair. % \textbf{Value}. The evaluation $|R|>0$ Judging the \textbf{value} of the result $R$ may be carried out ``locally'' (as an embedded part of the process of invention of $R$) or ``globally'' (i.e.~as an external process). % \textbf{Dynamic world}. As noted above, $T$ (and $T^\star$) appears within a stream of data with indeterminacy. There is a further feedback loop, insofar as products $R$ influence the future state. state of the system. Thus, the model exists in a \textbf{dynamic world}. % \textbf{Multiple contexts}. This is reflected directly in our Our model by the difference between separates the ``context of discovery'' discovery'', involving prior preparations $p$, and from the ``context of invention'' involving prior preparations $p^{\prime}$. Both of these may be subdivided further. further into \textbf{multiple contexts}. % \textbf{Multiple tasks}. Both $T$ and $T^\star$ may be multiple, causing an individual process to fork intocommunicating sub-processes dealing with \textbf{multiple tasks} that involve different skills sets. % \textbf{Multiple influences}. The process as a whole may be multiplied out among different communicating investigators. investigators, so that the final result bears the mark of \textbf{multiple influences}.

\end{enumerate} \end{quote} This can be summarised schematically as follows: % \input{schematic-tikz} {\centering \includegraphics[width=.8\textwidth]{schematic}

required to evaluate their own results, we are also implicitly requiring them to evaluate their creative process. We should give them the tools to do that effectively. % These ideas set a relatively high bar, if only because computational creativity has often been focused on generative rather than reflective acts. As Campbell \citeyear{campbell} writes: ``serendipity

\subsection{A thought \subsection{Thought experiment evaluating our model of serendipity} \label{sec:ww} To evaluate our computational framework in usage, we apply a thought experiment based around a scenario where there is high potential for