Just a reminder, if you are interested in such things, that the 2018 Code{4}Lib Conference starts on February 13th and runs through the 16th.  The conference is being sponsored by the Library of Congress, and this year also by ESRI. Code{4}Lib (or Code for Libraries) is a technology and open source driven collective of hackers, designers, architects, curators, data visualization experts, GIS aficionados, artists and instigators from around the world, who largely work for and/or with libraries, archives and museums on technology and data "stuff." The conference brings together an amazing group of people to hear about and to present the latest research on coding, computational techniques, and big data, being applied in Museums, Archives and Libraries, around the world. The group also sponsors a free journal dedicated to coding and computation in libraries, museums and archives


The keynote address for the conference is being given by Chris Bourg, who is the Director of Libraries at MIT,  and who will be talking about new technologies that are beginning in influence libraries, museums and archives, and about her thoughts on some of the big data challenges facing information retrieval and distribution in the age of infinite information. The second keynote will be given by Mega Subramaniam, who is an associate professor of computer science and information studies and is the director of the Libraries Ready to Code Program at the ALA. Her research is on the use of school and public libraries as effective learning environments to encourage science, technology, engineering and mathematics (STEM) interest among underserved, underrepresented, and disadvantaged young adults.


I will be giving a paper, along with Laruen Dimonte, from the University of Rochester, and Nilesh Patil, from Google, who worked on using convolutional neural nets for galaxy classification, in a session on Deep Learning Techniques and Historical Collections …. This paper presents an introduction to and a possible algorithmic framework for the building of a feature extractor that employs large convolutional neural networks to identify and extract layer features from historical maps that could be used in environmental, urban planning and development studies. The mathematical analysis of deep convolutional neural networks for feature extraction was first initiated by S. Mallat, in a 2012 ground breaking paper entitled, Group Invariant Scattering.  The paper I am presenting extends Mallet’s theoretical framework, with insights from the work of Tishby on the information bottleneck, into the analysis of algorithms for layer extraction from historical maps and shows the huge potential for future applications of neural networks and deep learning methods in making available the largely untapped reservoirs of geospatial data in cartographic collections around the world.


This is an amazing conference that brings together some really brilliant, sometimes really strange, but all pushing the limits of technological applications in museums and libraries…it is a great way to see where the field is moving, to look at how data is changing what we do in Libraries, to take in some coding workshops, and of course, to spur your imagination.