Thursday, February 2, 2017

Tangents - Deep Learning Methodologies

I attended an extremely well presented talk on Generative Adversarial Networks (GANs) yesterday, an approach for deep learning in the area of machine learning and artificial intelligence. This area of science is of use to the computational anthropologist as a means of better understanding large data sets. These data sets include the mixed media of images, text, music and movies that are increasingly critical to understanding modern culture and communication. There is very impressive work in this area that could eventually open up these media to anthropological analysis.
 
The talk was given by Jennifer Sleeman from the University of Maryland, Baltimore Campus, at George Mason University's Foundation Hall and hosted by Data Science DC, https://www.meetup.com/Data-Science-DC/. Her presentation was an excellent introduction to Generative Adversarial Networks for the layman. The methodology was first introduced by Ian Goodfellow and his colleagues, http://papers.nips.cc/paper/5423-generative-adversarial-nets.pdf 


GANs are two neural networks, discriminatory and generative, that compete to produce ‘real’ images. They are set up in a game theory dynamic and play a minimax zero-sum game. The Discriminator, starting with real training images, maps input images from the generator network to a desired output and is provides probabilistic feedback. The Generator, starting with an image of random noise, creates an image and compares it with feedback from the discriminator. Over time the Generator produces better images based on the feedback. In effect the Discriminator trains the Generator to produce the best image. A nice example of this can be found here: https://oshearesearch.com/index.php/2016/07/01/mnist-generative-adversarial-model-in-keras/

https://oshearesearch.com/wp-content/uploads/2016/07/mnist_gan.png
GAN Architecture, Source: O'Shea Research



I am struck by how creatively this methodology utilizes game theory to improve the machine learning process in neural networks. It is also an impressive example of how computers can learn and provides us with a model of how learning can occur in humans. Humans use creativity and innovation to produce new ideas and devices, and then we compare it to our concrete knowledge and understanding of the world. Over time the continuous generation of new ideas leads to novel human behavior that changes culture.

GANs can currently be used to classify images and create new data sets. It is limited to smaller image resolutions, 64x64, and images, and has problems with animal body parts, 3-D images and global structure. However work in this area continues with application to text and music data. 

More example code:



Useful Python libraries for deep learning and GANs include Keras, Torch, and TensorFlow.

Finally, for those interested in getting up to speed on GANs, here is a link to a free online book that provides the basics on deep learning: http://www.deeplearningbook.org/
 

Friday, January 6, 2017

Beginnings...


With this posting I begin the real work of my dissertation research project. I have successfully defended my topic proposal and can finally implement my own research plan. It has already been a long journey, and there is much more to come. Along the way I will share the ups and downs of this research project, interesting findings, and the many lessons, but today I introduce my dissertation project, From Networks to Recovery: Effects of social networks on community recovery in the face of flooding disasters.

How do social networks influence individual and community behavior in disasters? Because of the increasing frequency of flooding events and the large proportion of populations that are impacted by these events, many communities will struggle to survive, recover and thrive. Research into how communities nurture and leverage their social capital, i.e. social network connections, to obtain information and resources in times of stress will improve understanding of community resilience. In future I hope this understanding will help community leaders and decision-makers develop better policies to support the networks and behaviors that improve community resilience.

What are the significant underlying factors that impact social network change and utilization during disaster recovery, and what sets of variables and interactions in these networks lead to recovery? How much impact does space have on social networks? To answer these questions I plan to develop an agent-based model of social networks grounded in space to study the ability of these communities to rebuild their social networks and recover resources in the context of flooding. Once developed, I will use the model to test how variations in flooding impact, urban density, population characteristics and network forms impact the disaster recovery.

By using a holistic approach, integrating ethnographic research and disaster theory with computational experimentation, I intend to provide an example of how local, individual relationships aggregate into collective action and interact with global processes. My work will inevitably be limited by the quantity and quality of empirical data that directly apply to the research questions. Of practical necessity, the model will be a simplification of real-world behavior and event outcomes. Nevertheless computational methods will allow me to monitor and trace a subset of the variables interacting in the complex adaptive system of a community in disaster.

I am excited to begin this stage of the PhD process, and I welcome you to join me!