a high-throughput screening approach to discovering good forms of biologically inspired visual representation高通量筛选的方法来发现良好的生物启发可视化表示形式.pdfVIP

A High-Throughput Screening Approach to Discovering Good Forms of Biologically Inspired Visual Representation Nicolas Pinto1,2*, David Doukhan1,2, James J. DiCarlo1,2, David D. Cox1,2,3* 1 McGovern Institute for Brain Research, Massachusetts Institute of Technology, Cambridge, Massachussetts, United States of America, 2 Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, Massachussetts, United States of America, 3 The Rowland Institute at Harvard, Harvard University, Cambridge, Massachusetts, United States of America Abstract While many models of biological object recognition share a common set of ‘‘broad-stroke’’ properties, the performance of any one model depends strongly on the choice of parameters in a particular instantiation of that model—e.g., the number of units per layer, the size of pooling kernels, exponents in normalization operations, etc. Since the number of such parameters (explicit or implicit) is typically large and the computational cost of evaluating one particular parameter set is high, the space of possible model instantiations goes largely unexplored. Thus, when a model fails to approach the abilities of biological visual systems, we are left uncertain whether this failure is because we are missing a fundamental idea or because the correct ‘‘parts’’ have not been tuned correctly, assembled at sufficient scale, or provided with enough training. Here, we present a high-throughput approach to the exploration of such parameter sets, leveraging recent advances in stream processing hardware (high-end NVIDIA graphic cards and the PlayStation 3’s IBM Cell Processor). In analogy to high- throughput screening approaches in molecular biology and genetics, we explored thousands of potential network architectures and parameter instantiation