Exploring the Challenges towards Lifelong Fact Learning

By Mohamed Elhoseiny, Francesca Babiloni, Rahaf Aljundi, Marcus Rohrbach, Manohar Paluri, Tinne Tuytelaars. 2018

12 citations recorded by the source corpus.

So far life-long learning (LLL) has been studied in relatively small-scale and relatively artificial setups. Here, we introduce a new large-scale alternative. What makes the proposed setup more natural and closer to human-like visual systems is threefold: First, we focus on concepts (or facts, as we call them) of varying complexity, ranging from single objects to more complex structures such as objects performing actions, and objects interacting with other objects. Second, as in real-world settings, our setup has a long-tail distribution, an aspect which has mostly been ignored in the LLL context. Third, facts across tasks may share structure (e.g., \(\langle \)person, riding, wave\(\rangle \) and \(\langle \)dog, riding, wave\(\rangle \)). Facts can also be semantically related (e.g., “liger” relates to seen categories like “tiger” and “lion”). Given the large number of possible facts, a LLL setup seems a natural choice. To avoid model size growing over time and to optimally exploit the semantic relations and structure, we combine it with a visual semantic embedding instead of discrete class labels. We adapt existing datasets with the properties mentioned above into new benchmarks, by dividing them semantically or randomly into disjoint tasks. This leads to two large-scale benchmarks with 906,232 images and 165,150 unique facts, on which we evaluate and analyze state-of-the-art LLL methods.

Subfields: Meta & Continual Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Meta & Continual Learning

Citations in the corpus

LAMOL: LAnguage MOdeling for Lifelong Language Learning (2019, methodology): “Additionally, experiments such as Elhoseiny et al. (2018) show that MAS has better performance overall.”
Task-Free Continual Learning (2018, background): “Alternatively, a projection into an embedding space could be used, as in [7], avoiding the need for a growing network architecture.”
Compositional Continual Language Learning (2020, unknown): “(Rebuffi et al., 2017; Lopez-Paz & Ranzato, 2017; Shin et al., 2017; Li & Hoiem, 2016; Shmelkov et al., 2017; Triki et al., 2017; Li & Hoiem, 2016; Triki et al., 2017; Rusu et al., 2016; Lee et al., 2017; Elhoseiny et al., 2018; Kirkpatrick et al., 2017c; Zenke et al., 2017; Chaudhry et al., 2018)).”
Continual Learning in Neural Networks (2019, background): “This work was published as a conference paper at ACCV2018 [34] but not detailed in this manuscript.”
Normalization Matters in Zero-Shot Learning (2020, methodology): “[15] Mohamed Elhoseiny, Francesca Babiloni, Rahaf Aljundi, Marcus Rohrbach, Manohar Paluri, and Tinne Tuytelaars.”
LAMAL: LAnguage Modeling Is All You Need for Lifelong Language Learning (2019, methodology): “Additionally, some experiments such as (Elhoseiny et al. 2018) show that MAS has better performance overall.”
Long Live the Lottery: The Existence of Winning Tickets in Lifelong Learning (2021, background): “This paper’s study subject is class-incremental learning (CIL) (Rebuffi et al., 2017; Elhoseiny et al., 2018): a popular, realistic, albeit challenging setting of lifelong learning.”