BWhat changed
Reinterpreted Inception modules through depthwise-separable convolutions and proposed Xception, reporting image-classification gains at similar parameter count to Inception V3.
Scientific scope
Reinterpreted Inception modules through depthwise-separable convolutions and proposed Xception, reporting image-classification gains at similar parameter count to Inception V3. Scope limitation: First preprint was in 2016; year follows CVPR 2017 publication. The paper does not originate depthwise-separable convolution.
Evidence
The retained primary-source abstract, bibliographic record and stated review scope support this provisional assessment. No independent experimental replication by WRS is claimed. First preprint was in 2016; year follows CVPR 2017 publication. The paper does not originate depthwise-separable convolution.
Contributors
- The named authors listed in the linked research record · SCIENCE: Complete source byline retained in the research record; human research direction, method development and evaluation. Individual labor shares are not measured.
- Chollet: Xception depthwise-separable convolutional architecture — original paper
- Chollet: Xception depthwise-separable convolutional architecture — company research source
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction