The Odd One Out test of intelligence consists of 3x3 matrix reasoning problems organized in 20 levels of difficulty. Addressing problems on this test appears to require integration of multiple cognitive abilities usually associated with creativity, including visual encoding, similarity assessment, pattern detection, and analogical transfer. We describe a novel fractal strategy for addressing visual analogy problems on the Odd One Out test. In our strategy, the relationship between images is encoded fractally, capturing important aspects of similarity as well as inherent self-similarity. The strategy starts with fractal representations encoded at a high level of resolution, but, if that is not sufficient to resolve ambiguity, it automatically adjusts itself to the right level of resolution for addressing a given problem. Similarly, the strategy starts with searching for fractally-derived similarity between simpler relationships, but, if that is not sufficient to resolve ambiguity, it automatically shifts to search for such similarity between higher-order relationships. We present preliminary results and initial analysis from applying the fractal technique on nearly 3,000 problems from the Odd One Out test.
Fitzgerald, Tesca (Georgia Institute of Technology) | McGreggor, Keith (Georgia Institute of Technology) | Akgun, Baris (Georgia Institute of Technology) | Goel, Ashok K. (Georgia Institute of Technology) | Thomaz, Andrea L. (Georgia Institute of Technology)
Learning by observation is an important goal in developing complete intelligent robots that learn interactively. We present a visual analogy approach toward an integrated, intelligent system capable of learning skills from observation. In particular, we focus on the task of retrieving a previously acquired case similar to a new, observed skill. We describe three approaches to case retrieval: feature matching, feature transformation, and fractal analogy. SIFT features and fractal encoding were used to represent the visual state prior to the skill demonstration, the final state after the skill has been executed, and the visual transformation between the two states. We discovered that the three methods (feature matching, feature transformation, and fractal analogy) are useful for retrieval of similar skill cases under different conditions pertaining to the observed skills.
We report a novel approach to addressing the Raven’s Progressive Matrices (RPM) tests, one based upon purely visual representations. Our technique introduces the calculation of confidence in an answer and the automatic adjustment of level of resolution if that confidence is insufficient. We first describe the nature of the visual analogies found on the RPM. We then exhibit our algorithm and work through a detailed example. Finally, we present the performance of our algorithm on the four major variants of the RPM tests, illustrating the impact of confidence. This is the first such account of any computational model against the entirety of the Raven’s.
In the past, many knowledge representation systems failed because they were too monolithic and didn't scale well, whereas other systems failed to have an impact because they were small and isolated. Along with this trade-off in size, there is also a constant tension between the cost involved in building a larger community that can interoperate through common terms and the cost of the lack of interoperability. Its main contribution is in recognizing and supporting the fractal patterns of scalable web systems. In this article we discuss why fractal patterns are an appropriate model for web systems and how semantic web technologies can be used to design scalable and interoperable systems.
Alex P. Pentland Artificial Intelligence Center, SRI International 333 Ravenswood Ave., Menlo Park, California 94025 ABSTRACT Shape-from-shading and shape-from-texture methods have the To accomplish this, we must have rccour8e to a 3-D model competent to describe both crumpled surface8 and smooth ones. The fractal model of surface shape [6,7] appears to possess the required properties. Evidence for this comes from recently conducted surveys of natural imagery [6,8]. These survey found that the fractal model of imaged 3-D surfaces furnishes an accurate description of most textured and shaded image regions. Perhaps even more convincing, however, is the fact that fractals look like natural surfaces [9,10,11].