diff --git a/.binder/postBuild b/.binder/postBuild
index de5a232c..bdd916dc 100644
--- a/.binder/postBuild
+++ b/.binder/postBuild
@@ -1,5 +1,7 @@
 #!/bin/bash
 
+ROOTDIR=$(git rev-parse --show-toplevel 2>/dev/null)
+
 git clone https://github.com/derlin/hepqpr-qallse
-cd hepqpr-qallse/src
-python setup.py install
+cp -r hepqpr-qallse/src/hepqpr $ROOTDIR/source
+ln -s $ROOTDIR/source/hepqpr $ROOTDIR/source/ja/hepqpr
diff --git a/.binder/requirements.txt b/.binder/requirements.txt
index bbd44c43..bfa69694 100644
--- a/.binder/requirements.txt
+++ b/.binder/requirements.txt
@@ -1,8 +1,13 @@
-qiskit
-qiskit-optimization
-qiskit-machine-learning
+qiskit==0.42.1
+qiskit-ibm-runtime==0.9.1
+qiskit-ibm-provider==0.4.0
+qiskit-optimization==0.5.0
+qiskit-machine-learning==0.6.0
+qiskit-experiments==0.5.0
+qiskit-nature==0.5.2
 matplotlib
 pylatexenc
+plotly
 pandas
 scikit-learn
 tabulate
diff --git a/source/ja/qkc_machine_learning.md b/source/ja/qkc_machine_learning.md
index 1b8a9816..4054a2c8 100644
--- a/source/ja/qkc_machine_learning.md
+++ b/source/ja/qkc_machine_learning.md
@@ -408,7 +408,7 @@ sampler = Sampler()
 first_two_inputs = np.concatenate(norm_train_data[:2]).flatten()
 
 job = sampler.run(manual_kernel, parameter_values=first_two_inputs, shots=10000)
-# quasi_dists[0]がmanual_kernelの測定結果のcountsｋら
+# quasi_dists[0]がmanual_kernelの測定結果のcountsから推定される確率分布
 fidelity = job.result().quasi_dists[0].get(0, 0.)
 print(f'|<φ(x_0)|φ(x_1)>|^2 = {fidelity}')
 ```