How Deep is the Feature Analysis underlying Rapid Visual Categorization?
Authors: Sven Eberhardt, Jonah G. Cader, Thomas Serre
NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We have conducted a large-scale psychophysics study to assess the correlation between computational models and human behavioral responses on a rapid animal vs. non-animal categorization task. We considered visual representations of varying complexity by analyzing the output of different stages of processing in three stateof-the-art deep networks. We found that recognition accuracy increases with higher stages of visual processing (higher level stages indeed outperforming human participants on the same task) but that human decisions agree best with predictions from intermediate stages. Overall, these results suggest that human participants may rely on visual features of intermediate complexity and that the complexity of visual representations afforded by modern deep network models may exceed the complexity of those used by human participants during rapid categorization. |
| Researcher Affiliation | Academia | Sven Eberhardt Jonah Cader Thomas Serre Department of Cognitive Linguistic & Psychological Sciences Brown Institute for Brain Sciences Brown University Providence, RI 02818 {sven2,jonah_cader,thomas_serre}@brown.edu |
| Pseudocode | No | The paper does not contain any pseudocode or clearly labeled algorithm blocks. |
| Open Source Code | No | The paper does not provide any specific links or explicit statements about the release of source code for the described methodology. |
| Open Datasets | Yes | A large set of (target) animal and (distractor) non-animal stimuli was created by sampling images from Image Net [4]. |
| Dataset Splits | No | While cross-validation was used for hyperparameter optimization of the SVM ('C regularization parameter optimized by cross-validation'), the paper does not specify a distinct validation dataset split with percentages or counts for the primary experiments or the overall dataset. The models were pre-trained on ImageNet, which has standard splits, but no specific validation set for the task at hand is detailed. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., GPU/CPU models, memory) used for running the experiments. |
| Software Dependencies | No | The paper mentions software like 'Caffe implementation [10]', 'linear SVM (scikit-learn [14] implementation)', and 'psi Turk framework [13] combined with custom javascript functions', but it does not specify any version numbers for these components. |
| Experiment Setup | No | The paper does not provide specific experimental setup details such as hyperparameter values (e.g., learning rate, batch size, epochs) or detailed training configurations for the computational models. |