Skip to content

update #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Aug 19, 2019
Merged

update #2

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
c2bccfc
upload the slide
wenzheli Aug 2, 2019
a139e82
upload the slide
wenzheli Aug 6, 2019
2476118
push the slides
wenzheli Aug 11, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
229 changes: 229 additions & 0 deletions 10. TopicModels/LDA_tutorial.ipynb
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,229 @@
{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![](attachment:image.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"![](attachment:image.png)"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": []
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Gibbs sampler for LDA"
]
},
{
"cell_type": "code",
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
"# words\n",
"import numpy as np\n",
"W = np.array([0, 1, 2, 3, 4])\n",
"\n",
"# D := document words\n",
"X = np.array([\n",
" [0, 0, 1, 2, 2],\n",
" [0, 0, 1, 1, 1],\n",
" [0, 1, 2, 2, 2],\n",
" [4, 4, 4, 4, 4],\n",
" [3, 3, 4, 4, 4],\n",
" [3, 4, 4, 4, 4]\n",
"])\n",
"\n",
"N_D = X.shape[0] # num of docs\n",
"N_V = W.shape[0] # num of words\n",
"N_K = 2 # num of topics"
]
},
{
"cell_type": "code",
"execution_count": 5,
"metadata": {},
"outputs": [],
"source": [
"import numpy as np\n",
"# Dirichlet priors\n",
"alpha = 1\n",
"gamma = 1\n",
"\n",
"# Z := word topic assignment\n",
"Z = np.zeros(shape=[N_D, N_V])\n",
"\n",
"for i in range(N_D):\n",
" for l in range(N_V):\n",
" Z[i, l] = np.random.randint(N_K) # randomly assign word's topic\n",
"\n",
"# Pi := document topic distribution\n",
"theta = np.zeros([N_D, N_K])\n",
"\n",
"for i in range(N_D):\n",
" theta[i] = np.random.dirichlet(alpha*np.ones(N_K))\n",
"\n",
"# phi := word topic distribution\n",
"phi = np.zeros([N_K, N_V])\n",
"\n",
"for k in range(N_K):\n",
" phi[k] = np.random.dirichlet(gamma*np.ones(N_V))\n",
"\n",
"for it in range(1000):\n",
" # Sample from full conditional of Z\n",
" # ---------------------------------\n",
" for i in range(N_D):\n",
" for v in range(N_V):\n",
" # Calculate params for Z\n",
" p_iv = np.exp(np.log(theta[i]) + np.log(phi[:, X[i, v]]))\n",
" p_iv /= np.sum(p_iv)\n",
"\n",
" # Resample word topic assignment Z\n",
" Z[i, v] = np.random.multinomial(1, p_iv).argmax()\n",
"\n",
" # Sample from full conditional of \\theta\n",
" # ----------------------------------\n",
" for i in range(N_D):\n",
" m = np.zeros(N_K)\n",
"\n",
" # Gather sufficient statistics\n",
" for k in range(N_K):\n",
" m[k] = np.sum(Z[i] == k)\n",
"\n",
" # Resample doc topic dist.\n",
" theta[i, :] = np.random.dirichlet(alpha + m)\n",
"\n",
" # Sample from full conditional of \\phi\n",
" # ---------------------------------\n",
" for k in range(N_K):\n",
" n = np.zeros(N_V)\n",
"\n",
" # Gather sufficient statistics\n",
" for v in range(N_V):\n",
" for i in range(N_D):\n",
" for l in range(N_V):\n",
" n[v] += (X[i, l] == v) and (Z[i, l] == k)\n",
"\n",
" # Resample word topic dist.\n",
" phi[k, :] = np.random.dirichlet(gamma + n)"
]
},
{
"cell_type": "code",
"execution_count": 6,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[[0.33941922 0.66058078]\n",
" [0.12011984 0.87988016]\n",
" [0.02318953 0.97681047]\n",
" [0.80301107 0.19698893]\n",
" [0.94780545 0.05219455]\n",
" [0.84677554 0.15322446]]\n"
]
}
],
"source": [
"print (theta)"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"### Mini-homework - Collapsed Gibbs Sampler for LDA\n",
"自己重新推导一下Collapsed Gibbs Sampler，并实现一下LDA类并利用上面给定的YELP数据集来测试一下，并跟sklearn的结果比较一下。"
]
},
{
"cell_type": "code",
"execution_count": 7,
"metadata": {},
"outputs": [
{
"ename": "IndentationError",
"evalue": "expected an indented block (<ipython-input-7-fe62b0541667>, line 7)",
"output_type": "error",
"traceback": [
"\u001b[0;36m File \u001b[0;32m\"<ipython-input-7-fe62b0541667>\"\u001b[0;36m, line \u001b[0;32m7\u001b[0m\n\u001b[0;31m def fit(self, X, y=None):\u001b[0m\n\u001b[0m ^\u001b[0m\n\u001b[0;31mIndentationError\u001b[0m\u001b[0;31m:\u001b[0m expected an indented block\n"
]
}
],
"source": [
"class LDA:\n",
" \"\"\"Latent Dirichlet allocation using collapsed Gibbs sampling\"\"\"\n",
" \n",
" def __init__():\n",
" \n",
"\n",
" def fit(self, X, y=None):\n",
" \"\"\"Fit the model with X.\"\"\"\n",
" \n",
"\n",
" def fit_transform(self, X, y=None):\n",
" \n",
"\n",
" def transform(self, X, max_iter=20, tol=1e-16):\n",
" \n",
" \n",
" def loglikelihood(self):\n",
" \"\"\"Calculate complete log likelihood, log p(w,z)\n",
" Formula used is log p(w,z) = log p(w|z) + log p(z)\n",
" \"\"\"\n",
" def perplexity(self):\n",
" \"\"\"Calculate the perplexity\"\"\"\n",
" "
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
}
],
"metadata": {
"kernelspec": {
"display_name": "Python 3",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.7.1"
}
},
"nbformat": 4,
"nbformat_minor": 2
}
Loading