hao

IJCAI2021/tex/introduction.tex

@@ -7,7 +7,7 @@ including philology, biology, and computer
 science. Especially in computer science, efforts in recent years trying to explore
 the emergent language in virtual multi-agent environments, where
 agents are trained to communicate with neural-network-based methods such as deep
-reinforcement learning~\cite{kottur-etal-2017-natural,bogin2018emergence,lazaridou2018emergence,choi2018compositional,jaques2019social,mul2019mastering,kharitonov2019egg,labash2020perspective,chaabouni2020compositionality}.
+reinforcement learning~\cite{kottur-etal-2017-natural,bogin2018emergence,lazaridou2018emergence,choi2018compositional,jaques2019social,mul2019mastering,kharitonov2019egg,labash2020perspective,chaabouni2020compositionality, gupta2020compositionality}.
 %Such works can be roughly classified into two categories,
 %referential game~\cite{} and multi-agent reinforcement learning (MARL)~\cite{}, based on
 %the environment setting.  
@@ -19,8 +19,8 @@ Compositionality is a principle that determines
 whether the meaning of a complex expression (e.g, phrase), which is assembled out of a
 given set of simple components (e.g., symbols), can be determined by its
 constituent components and the rule combining them~\cite{andreas2018measuring,chaabouni2020compositionality}.
-\note{For example, the expression ``AAAI is a conference'' consists of two
-meaningful words ``AAAI'' and ``conference'', and a rule for definition (``is'').
+\note{For example, the expression ``IJCAI is a conference'' consists of two
+meaningful words ``IJCAI'' and ``conference'', and a rule for definition (``is'').
 Compositionality is considered to be a source of productivity,
 systematicity, and learnability of language, and the reason why a language with finite
 vocabulary can express almost infinite concepts.}
@@ -66,6 +66,7 @@ vocabulary can express almost infinite concepts.}
 %    \cite{li2019ease}&Population size, resetting all listeners&Quantitative, Speaker\\
 %    \cite{chaabouni-etal-2019-word}&Word-order constraints&Not quantitative, Speaker\\
 %    \cite{chaabouni2020compositionality}&Easier to decode&Quantitative, Speaker\\
+%    \cite{gupta2020compositionality}&structural dependent models&Quantitative, Speaker\\
 %    \textbf{Ours} & \textbf{None} & \textbf{Quantitative, Speaker+Listener} \\
 %    \bottomrule
 %    \end{tabular}
@@ -74,7 +75,7 @@ vocabulary can express almost infinite concepts.}
 Prior studies focus on achieving high compositional symbolic language
 through \emph{deliberately handcrafted} inductions, e.g., additional rewards~\cite{mordatch2017emergence},
 constructed loss functions~\cite{kharitonov2019egg}, structural input data~\cite{lazaridou2018emergence,evtimova2018emergent},
-memoryless~\cite{kottur-etal-2017-natural,li2019ease}, and ease-of-teaching~\cite{li2019ease}.
+memoryless~\cite{kottur-etal-2017-natural,li2019ease}, ease-of-teaching~\cite{li2019ease}, and structural dependent models~\cite{gupta2020compositionality}.
 \note{Such optimization methodologies are driven by the challenges to generate high compositional symbolic without induction in an existing multi-agent environment.}
 
 Figure~\ref{fig:induction} reports the compositionality when training two agents

IJCAI2021/tex/relatedwork.tex

@@ -7,7 +7,7 @@ Previous works focus on the \emph{deliberately handcrafted} inductions that affe
 compositionality of emergent language.
 Some significant works on studying the environmental inductions on the compositionality of emergent language are summarized in Table~\ref{tab:rel}.
 For example, 
-~\cite{gupta2020compositionality} used a structural dependent agents, i.e., a variational autoencoder, to simulate a speaker-listener pair, and constrained the communication channel into a binary coding sequence.
+~\cite{gupta2020compositionality} used a structural dependent models, i.e., a variational autoencoder, to simulate a speaker-listener pair, and constrained the communication channel into a binary coding sequence.
 ~\cite{kirby2015compression} explored how the pressures for expressivity and compressibility lead the structured language.
 ~\cite{kottur-etal-2017-natural} constrained the vocabulary size and whether the listener has memory to coax the compositionality of the emergent language.
 ~\cite{lazaridou2018emergence} showed that the degree of structure found in the input data affects the emergence of the symbolic language.