Commit 4302acd8 by Zidong Du
parents 3a7a4d87 8de2e156
......@@ -18,14 +18,42 @@
\label{fig:exp2}
\end{figure}
\begin{figure}[t]
%\begin{figure}[t]
% \centering
% \includegraphics[width=0.99\columnwidth]{fig/Figure10_p_value.pdf}
% \caption{The Chi-square test between high-compositionality and agent
% capacity. (a) $MIS>0.99$. (b)
% $MIS>0.9$.}
% \label{fig:exp10}
%\end{figure}
\begin{table}[b]
\centering
\includegraphics[width=0.99\columnwidth]{fig/Figure10_p_value.pdf}
\caption{The Chi-square test between high-compositionality and agent
capacity. (a) $MIS>0.99$. (b)
$MIS>0.9$.}
\label{fig:exp10}
\end{figure}
\caption{The Chi-square test between high-compositionality and agent capacity.}
\label{tab:exp10}
\begin{tabular}{cccc}
\toprule
\multicolumn{4}{c}{$H_0$: $\mathit{MIS} > 0.90$ is independent with $h_{\mathit{size}}$}\\
\midrule
Vocabulary size & $\chi^2$ & $df$ & $p$-value \\
\midrule
4 & 22.20 & 10 & $1.41\times 10^{-2}$ \\
6 & 27.52 & 10 & $2.16\times 10^{-3}$ \\
10 & 64.46 & 10 & $5.14\times 10^{-10}$ \\
\bottomrule
\multicolumn{4}{c}{\vspace{1em}}\\
\toprule
\multicolumn{4}{c}{$H_0$: $\mathit{MIS} > 0.99$ is independent with $h_{\mathit{size}}$}\\
\midrule
Vocabulary size & $\chi^2$ & $df$ & $p$-value \\
\midrule
4 & 30.19 & 10 & $7.97\times 10^{-4}$ \\
6 & 25.96 & 10 & $3.80\times 10^{-3}$ \\
10 & 33.80 & 10 & $2.00\times 10^{-4}$ \\
\bottomrule
\end{tabular}
\end{table}
\begin{figure}[t]
\centering
......@@ -84,8 +112,8 @@ more meanings, for the constraint from low capacity.
Additionally, we also perform $\chi^2$ test to check the statistical
significance between the high compositionality and agent
capacity. Figure~\ref{fig:exp10} reports the $\chi^2$ test results for
$MIS>0.99$ and $MIS>0.9$ in (a) and (b), respectively. It can be observed that
capacity. Table~\ref{tab:exp10} reports the $\chi^2$ test results for
$\mathit{MIS}>0.99$ and $\mathit{MIS}>0.9$, respectively. It can be observed that
for different vocabulary size, the p-value is always less than 0.05, which means
the high compositionality has statistical significance related to agent
capacity.
......@@ -99,7 +127,7 @@ capacity.
\begin{figure*}[t]
\centering
\includegraphics[width=1.8\columnwidth]{fig/Figure9.pdf}
\includegraphics[width=\textwidth]{fig/Figure9.pdf}
\caption{Accuracy of Listeners when varying $h_{size}$ from 1 to 8. Each curve
represents an average accuracy trend from 50 repeated training, with the
range of [$\mu - \sigma$, $\mu + \sigma$], where $\mu$ is the average
......
......@@ -5,14 +5,15 @@
%external environmental factors
Previous works focus on the external environmental factors that impact the
compositionality of emerged symbolic language.
Some significant works on studying the external environmental factor on the compositionality of emergent language are summarized on Table~\ref{tab:rel}.
For example, ~\citet{kirby2015compression} explored how the pressures for expressivity and compressibility lead the structured language.
~\citet{kottur-etal-2017-natural} constrained the vocabulary size and whether the listener has memory to coax the compositionality of the emergent language.
~\citet{lazaridou2018emergence} showed that the degree of structure found in the input data affects the emergence of the symbolic language.
~\citet{li2019ease} studied how the pressure, ease of teaching, impact on the iterative language of the population regime.
~\citet{evtimova2018emergent} designed a novel multi-modal scenarios, which the speaker and the listener should access to different modalities of the input object, to explore the language emergence.
Such factors are deliberately designed, which are too ideal to be true in
the real world. None of these works realizes the importance of model capacity of
agent itself. \rmk{this should be largely emphasized.}
the real world.
In this paper, these handcrafted inductions above are all removed, and the high compostional language is leaded only by the agent capacity.
......
......@@ -38,8 +38,8 @@ to finish the game in a cooperative manner. In each round, once received an
input object $t$, Speaker $S$ speaks a symbol sequence $s$ to Listener $L$ ;
Listener $L$ reconstruct the predicted result $\hat{t}$ based on the listened
sequence $s$; if $t=\hat{t}$, agents win this game and receive positive rewards
($R(t,\hat{t})=1$); otherwise agents fail this game and receive negative rewards
($R(t,\hat{t})=-1$).
($r(t,\hat{t})=1$); otherwise agents fail this game and receive negative rewards
($r(t,\hat{t})=-1$).
Precisely, during the game, Speaker $S$ receives an input object $t$, which is
an expression with two words from the vocabulary set $V$, i.e., two
......@@ -87,7 +87,7 @@ Algorithm~\ref{al:learning}, we train the separate Speaker $S$ and Listener $L$
Stochastic Policy Gradient methodology in a tick-tock manner, i.e, training one
agent while keeping the other one. Roughly, when training the Speaker, the
target is set to maximize the expected reward
$J(\theta_S, \theta_L)=E_{\pi_S,\pi_L}[R(t, \hat{t})]$ by adjusting the parameter
$J(\theta_S, \theta_L)=E_{\pi_S,\pi_L}[r(t, \hat{t})]$ by adjusting the parameter
$\theta_S$, where $\theta_S$ is the neural network parameters of Speaker $S$
with learned output probability distribution $\pi_S$, and $\theta_L$ is the
neural network parameters of Listener with learned probability distribution $\pi_L$.
......@@ -100,9 +100,9 @@ use the predict result $\hat{t}$ of the listener agent as the
evidence of whether giving the positive rewards. Then, the gradients of the
expected reward $ J(\theta_S, \theta_L)$ can be calculated as follows:
\begin{align}
\nabla_{\theta^S} J &= \mathbb{E}_{\pi^S, \pi^L} \left[ R(\hat{t}, t) \cdot
\nabla_{\theta^S} J &= \mathbb{E}_{\pi^S, \pi^L} \left[ r(\hat{t}, t) \cdot
\nabla_{\theta^S} \log{\pi^S(s_0, s_1 | t)} \right] \\
\nabla_{\theta^L} J &= \mathbb{E}_{\pi^S, \pi^L} \left[ R(\hat{t}, t) \cdot
\nabla_{\theta^L} J &= \mathbb{E}_{\pi^S, \pi^L} \left[ r(\hat{t}, t) \cdot
\nabla_{\theta^L} \log{\pi^S(\hat{t} | s_0, s_1)} \right]
\end{align}
......@@ -119,8 +119,8 @@ expected reward $ J(\theta_S, \theta_L)$ can be calculated as follows:
\STATE Sample $s_0$ with $P(s_0|t)$, $s_1$ with $P(s_1|t)$
\STATE $P(\hat{t}|s) = \pi^L(\hat{t}|s)$
\STATE Sample $\hat{t}$ with $P(\hat{t}|s)$
\STATE Get reward $R(\hat{t},t)$
\STATE $J(\theta^S,\theta^L)=E_{\pi_{old}^S,\pi^L}[R(\hat{t},t)\cdot\frac{\pi^S(s|t)}{\pi^S_{old}(s|t)}]$
\STATE Get reward $r(\hat{t},t)$
\STATE $J(\theta^S,\theta^L)=E_{\pi_{old}^S,\pi^L}[r(\hat{t},t)\cdot\frac{\pi^S(s|t)}{\pi^S_{old}(s|t)}]$
\STATE Update $\theta^S$ by $\bigtriangledown_{\theta^S}J$
\ENDFOR
\STATE $\pi_{old}^S\leftarrow \pi^S$
......@@ -134,8 +134,8 @@ expected reward $ J(\theta_S, \theta_L)$ can be calculated as follows:
\STATE Sample $s_0$ with $P(s_0|t)$, $s_1$ with $P(s_1|t)$
\STATE $P(\hat{t}|s) = \pi^L_{old}(\hat{t}|s)$
\STATE Sample $\hat{t}$ with $P(\hat{t}|s)$
\STATE Get reward $R(\hat{t},t)$
\STATE $J(\theta^S,\theta^L)=E_{\pi_{old}^S,\pi^L}[R(\hat{t},t)\cdot\frac{\pi^L(s|t)}{\pi^L_{old}(s|t)}]$
\STATE Get reward $r(\hat{t},t)$
\STATE $J(\theta^S,\theta^L)=E_{\pi_{old}^S,\pi^L}[r(\hat{t},t)\cdot\frac{\pi^L(s|t)}{\pi^L_{old}(s|t)}]$
\STATE Update $\theta^L$ by $\bigtriangledown_{\theta^L}J$
\ENDFOR
\STATE $\pi_{old}^L\leftarrow \pi^L$
......
......@@ -36,7 +36,7 @@ Each column of $M$ correspond to the semantic information carried by one symbol.
\begin{figure}[t]
\centering
\includegraphics[width=\columnwidth]{fig/Figure5_An_emergent_language.pdf}
\includegraphics[width=0.8\columnwidth]{fig/Figure5_An_emergent_language.pdf}
\caption{An emergent language that the unilateral metrics cannot measure its non-compositionality. Notice that given $s_1 = \mathrm{a}$, the listener can neither determine the shape nor the color without the knowledge about $s_0$.}
\label{fig:unilateral}
\end{figure}
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment