Title :

FiLM: Visual Reasoning with a General Conditioning Layer

Author(s) :

Perez, Ethan [Auteur]
Rice University [Houston]
Strub, Florian [Auteur]
Centre de Recherche en Informatique, Signal et Automatique de Lille - UMR 9189 [CRIStAL]
Sequential Learning [SEQUEL]
de Vries, Harm [Auteur]
Dumoulin, Vincent [Auteur]
Courville, Aaron [Auteur]

Conference title :

AAAI Conference on Artificial Intelligence

City :

New Orleans

Country :

Etats-Unis d'Amérique

Start date of the conference :

2018-02-02

HAL domain(s) :

Informatique [cs]/Réseau de neurones [cs.NE]
Informatique [cs]/Intelligence artificielle [cs.AI]

English abstract : [en]

We introduce a general-purpose conditioning method for neu-ral networks called FiLM: Feature-wise Linear Modulation. FiLM layers influence neural network computation via a simple , feature-wise affine transformation based ...
Show more >We introduce a general-purpose conditioning method for neu-ral 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.Show less >

Language :

Anglais

Peer reviewed article :

Oui

Audience :

Internationale

Popular science :

Non

Collections :

• Centre de Recherche en Informatique, Signal et Automatique de Lille (CRIStAL) - UMR 9189

Source :

Harvested from HAL