Phil 1.28.18


  • A full-throated defense of simulation from Joanna Bryson in Artificial Intelligence and Pro-Social Behaviour (pg 290)
    • The role of simulations in science has been at times confused, not only by occasional bad practice (as with any method), but also by claims by some of the method’s innovators that simulations were a “third way” to do science (after induction and deduction, Axelrod 1997 ). However, more recently a consensus has been reached that simulation and modelling more generally are indeed a part of ordinary science (Dunbar 2002 ; Kokko 2007 ; Seth et al. 2012). The part that they are is theory building. Every model is a theory—a very-well specified theory. In the case of simulations, the models are theories expressed in so much detail that their consequences can be checked by execution on a computer. Science requires two things: theories that explain the world, and data about the world which can be used to compare and validate the theories. A simulation provides no data about the world, but it can provide a great deal of ‘data’ about a theory. First, the very process of constructing a simulation can show that a theory is incoherent—internally contradictory, or incomplete, making no account for some part of the system intended to be explained (Axelrod 1997 ; Whitehouse et al. 2012 ). Secondly, modelling in general can show us a fuller range of consequences for a theory. This allows us to make specific, formal hypotheses about processes too complex to entirely conceptualise inside a single human brain (Dunbar 2002 ; Kokko 2007 ). The wide-spread acceptance of simulations as a part of the scientific method can be seen by their inclusion in the highest levels of academic publication, both in the leading general science journals and in the flagship journals for specific fields ranging from biology through political science. Fortunately, a theory expressed formally as a simulation can also be expressed in the traditional, informal, ordinary-language way as well.
  • Also this, from the same article:
    • Recently in the megafauna literature there has been a new hypothesis: individuals in populations might benefit from information transmission, of which vigilance against predators is just a special case (Crockford et al. 2012 ; Chivers and Ferrari 2014 ; Hogan and Laskowski 2013 ; Derex et al. 2013 ). Transmission of behaviour may be at least as important as information about localised threats (Jaeggi et al. 2008 ; Dimitriu et al. 2014 ). Note that behaviour itself, when transmitted horizontally (that is, not by genes to offspring), must be transmitted as information via perception (Shannon 2001 ).
  • On Discovering the Number of Document Topics via Conceptual Latent Space
    • Topic modeling is a widely used technique in knowledge discovery and data mining. However, finding the right number of topics in a given text source has remained a challenging issue. In this paper, we study the concept of conceptual stability via nonnegative matrix factorization. Based on this finding, we propose a method to identify the correct number of topics and offer empirical evidence in its favor in terms of classification accuracy and the number of topics that are naturally present in the text sources. Experiments on real-world text corpora demonstrate that the proposed method has outperformed state-of-the-art latent Dirichlet allocation and nonnegative matrix factorization models.
  • Beyond the Ranked List: User-Driven Exploration and Diversification of Social Recommendation
    • The beyond-relevance objectives of recommender systems have been drawing more and more attention. For example, a diversity-enhanced interface has been shown to associate positively with overall levels of user satisfaction. However, little is known about how users adopt diversity-enhanced interfaces to accomplish various real-world tasks. In this paper, we present two attempts at creating a visual diversity-enhanced interface that presents recommendations beyond a simple ranked list. Our goal was to design a recommender system interface to help users explore the different relevance prospects of recommended items in parallel and to stress their diversity. Two within-subject user studies in the context of social recommendation at academic conferences were conducted to compare our visual interfaces. Results from our user study show that the visual interfaces significantly reduced the exploration efforts required for given tasks and helped users to perceive the recommendation diversity. We show that the users examined a diverse set of recommended items while experiencing an improvement in overall user satisfaction. Also, the users’ subjective evaluations show significant improvement in many user-centric metrics. Experiences are discussed that shed light on avenues for future interface designs.
  • Setting up a Dissertation main points page on Phlog
  • This is an interesting map, from allgeneralizationsarefalse.commedia-bias-chart_3-0_hi-res
  • Don’t know what to do with this, but wow: Audio Adversarial Examples: Targeted Attacks on Speech-to-Text
    • We construct targeted audio adversarial examples on automatic speech recognition. Given any audio waveform, we can produce another that is over 99.9% similar, but transcribes as any phrase we choose (at a rate of up to 50 characters per second). We apply our iterative optimization-based attack to Mozilla’s implementation DeepSpeech end-to-end, and show it has a 100% success rate. The feasibility of this attack introduce a new domain to study adversarial examples.