仿射变换
计算机科学
水准点(测量)
人工智能
特征(语言学)
转化(遗传学)
计算
人工神经网络
简单(哲学)
图层(电子)
图像(数学)
任务(项目管理)
视觉推理
过程(计算)
模式识别(心理学)
机器学习
算法
数学
哲学
地理
纯数学
管理
化学
有机化学
经济
操作系统
认识论
基因
生物化学
语言学
大地测量学
作者
Ethan Perez,Florian Strub,Harm de Vries,Vincent Dumoulin,Aaron Courville
出处
期刊:Proceedings of the ... AAAI Conference on Artificial Intelligence
[Association for the Advancement of Artificial Intelligence (AAAI)]
日期:2018-04-29
卷期号:32 (1)
被引量:931
标识
DOI:10.1609/aaai.v32i1.11671
摘要
We introduce a general-purpose conditioning method for neural networks called FiLM: Feature-wise Linear Modulation. FiLM layers influence neural network computation via a simple, feature-wise affine transformation based on conditioning information. We show that FiLM layers are highly effective for visual reasoning - answering image-related questions which require a multi-step, high-level process - a task which has proven difficult for standard deep learning methods that do not explicitly model reasoning. Specifically, we show on visual reasoning tasks that FiLM layers 1) halve state-of-the-art error for the CLEVR benchmark, 2) modulate features in a coherent manner, 3) are robust to ablations and architectural modifications, and 4) generalize well to challenging, new data from few examples or even zero-shot.
科研通智能强力驱动
Strongly Powered by AbleSci AI