Emergence of Linguistic Communication from Referential Games with Symbolic and Pixel Input

By Angeliki Lazaridou, Karl Moritz Hermann, Karl Tuyls, Stephen Clark. 2018

247 citations recorded by the source corpus.

The ability of algorithms to evolve or learn (compositional) communication protocols has traditionally been studied in the language evolution literature through the use of emergent communication tasks. Here we scale up this research by using contemporary deep learning methods and by training reinforcement-learning neural network agents on referential communication games. We extend previous work, in which agents were trained in symbolic environments, by developing agents which are able to learn from raw pixel data, a more challenging and realistic input representation. We find that the degree of structure found in the input data affects the nature of the emerged protocols, and thereby corroborate the hypothesis that structured compositional language is most likely to emerge when agents perceive the world as being structured.

Subfields: Artificial Intelligence, Cultural Studies, Natural Language Processing

How researchers see this paper

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

In Artificial Intelligence

In Cultural Studies

In Natural Language Processing

Citations in the corpus

Experience Grounds Language (2020, background): “testbed for studying language (Werner and Dyer, 1991) and emergent communication (Schlangen, 2019a; Lazaridou et al., 2018; Chaabouni et al., 2020).”
XOC: Explainable Observer-Classifier for Explainable Binary Decisions (2019, background): “For instance, image reference games are used to study the emergence of language [22] and how agents can learn to communicate more effectively even when concepts are being misunderstood [6].”
ReferIt3D: Neural Listeners for Fine-Grained 3D Object Identification in Real-World Scenes (2020, background): “Recently, these approaches have also been adopted as a benchmark for discriminative and context-aware NLP [47,10,46,19,62,60,34,4].”
Compositionality and Generalization In Emergent Languages (2020, methodology): “Topographic similarity has become the standard way to quantify the compositionality of emergent languages (e.g., Brighton and Kirby, 2006; Lazaridou et al., 2018; Li and Bowling, 2019).”
Learning Decision Trees Recurrently Through Communication (2021, background): “Image reference games are used to study the emergence of language [37] and effectiveness in communication also when concepts are being misunderstood [9].”
Learning to Teach in Cooperative Multiagent Reinforcement Learning (2018, background): “ative MARL have shown that adding a communication mechanism between agents signiﬁcantly improves ﬁnal team performance in tasks (Foerster et al., 2016; Mordatch & Abbeel, 2017; Lowe et al., 2017; Lazaridou et al., 2018). Our work builds on these results by adopting communication mechanisms to en”
Multi-Agent Reinforcement Learning: A Selective Overview of Theories and Algorithms (2019, background): “Among these works, [346, 345, 348, 353, 354] have reported the emergence of computational communication protocols among the agents when the RL algorithm is trained from scratch with text or image inputs.”
Iterated learning for emergent systematicity in VQA (2021, background): “Several works in the literature have found that compositionality only arises in emergent languages if appropriate environmental pressures are present (Kottur et al., 2017; Choi et al., 2018; Lazaridou et al., 2018; Chaabouni et al., 2020).”
Measuring Compositionality in Representation Learning (2019, background): “…of evidence for compositionality—if the distance function rewards pairs of representations that share overlapping substructure (as might be the case with e.g. string edit distance), edit distance will be expected to correlate with some notion of derivational similarity (Lazaridou et al., 2018).”
Social Influence as Intrinsic Motivation for Multi-Agent Deep Reinforcement Learning (2018, background): “…emergent communication protocols also remains a challenging problem; recent empirical results underscore the difficulty of learning meaningful emergent communication protocols, even when relying on centralized training (e.g., Lazaridou et al. (2018); Cao et al. (2018); Foerster et al. (2016)).”
Pommerman: A Multi-Agent Playground (2018, background): “…learning (Foerster et al. 2016; Resnick et al. 2018d; Evtimova et al. 2017; Foerster et al. 2017; Lewis et al. 2017; Mordatch and Abbeel 2017; Lazaridou et al. 2018) with established prior work as well (Steels 1999; Steels 2003; Levy and Kirby 2006; Fehervari and Elmenreich 2010), while the…”
Multi-agent deep reinforcement learning: a survey (2021, background): “Many studies are concerned with the development of communication in referential games grounded on visual perception as it can be found in Choi et al. (2018), Evtimova et al. (2018), Havrylov and Titov (2017), Jorge et al. (2016), Lazaridou et al. (2018) and Lee et al. (2017).”
Dark, Beyond Deep: A Paradigm Shift to Cognitive AI with Humanlike Common Sense (2020, background): “The emergence of more realistic communication protocols using discrete messages has been explored in various types of communication games [445, 446, 447, 448], in which agents need to process visual signals and attach discrete tokens to at-”
Emergent Communication of Generalizations (2021, methodology): “The ability to communicate accurately, even on unseen concepts, is not necessarily indicative of more systematic communication; generalization without compositional language is a common ﬁnding in the literature [1, 6, 16, 22].”