Update

shashankkotyan · Feb 7, 2024 · 2b354fa · 2b354fa
1 parent 43012bd
commit 2b354fa
Show file tree

Hide file tree

Showing 7 changed files with 395 additions and 2 deletions.
diff --git a/README.md b/README.md
@@ -1,2 +1,32 @@
-# EvoSeed
-Source code for the article, "EvoSeed: Unveiling the Threat on Deep Neural Networks\\with Real-World Illusions"
+<div align="center">
+
+# **🎏 EvoSeed 🎏**
+
+</div>
+
+Source for the article: [EvoSeed: Unveiling the Threat on Deep Neural Networks with Real-World Illusions]()
+
+## 🌟 Features
+
+- A model-agnostic black-box algorithimic framework based on Evolutionary Strategy to generate unrestricted natural adversarial samples
+- A model-agnostic black-box algorithimic framework based on Evolutionary Strategy to generate unrestricted natural adversarial samples
+
+
+<div align="center">
+    <img src="./assets/compare.png" alt="" width="100%", style="background-color:white">
+</div>
+Visualization of the generated natural adversarial samples by Random Search (RandSeed) and proposed EvoSeed.
+
+## 📝 Citation
+
+If you find this project useful please cite:
+
+```
+@article{kotyan2024evoseed,
+    title = {EvoSeed: Unveiling the Threat on Deep Neural Networks with Real-World Illusions,
+    shorttitle = {EvoSeed},
+    author = {Kotyan, Shashank and Mao, PoYuan and Vargas, Danilo Vasconcellos},
+    year = {2024},
+    month = feb,
+}
+```
diff --git a/assets/compare.png b/assets/compare.png
diff --git a/assets/framework.png b/assets/framework.png
diff --git a/assets/illustration.png b/assets/illustration.png
diff --git a/assets/reference.bib b/assets/reference.bib
@@ -0,0 +1,7 @@
+@article{kotyan2024evoseed,
+    title = {EvoSeed: Unveiling the Threat on Deep Neural Networks with Real-World Illusions,
+    shorttitle = {EvoSeed},
+    author = {Kotyan, Shashank and Mao, PoYuan and Vargas, Danilo Vasconcellos},
+    year = {2024},
+    month = feb,
+}
diff --git a/index.html b/index.html
@@ -0,0 +1,139 @@
+<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
+<html xmlns="http://www.w3.org/1999/xhtml">
+
+<head>
+    <!-- ***** -->
+    <title>
+        EvoSeed: Unveiling the Threat on Deep Neural Networks with Real-World Illusions
+    </title>
+    <meta name="description"
+        content="Project page for &#39;EvoSeed: Unveiling the Threat on Deep Neural Networks with Real-World Illusions.&#39;">
+    <!-- ***** -->
+
+    <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
+
+    <style>
+        @import url('https://fonts.cdnfonts.com/css/chalkduster');
+    </style>
+
+    <meta name="viewport" content="width=device-width, initial-scale=1">
+
+    <link rel="stylesheet" href="style.css" type="text/css">
+    <link rel="stylesheet" href="https://fonts.cdnfonts.com/css/chalkduster" >
+    <link rel="stylesheet" href="https://cdn.jsdelivr.net/npm/bootstrap@5.2.3/dist/css/bootstrap.min.css" integrity="sha384-rbsA2VBKQhggwzxH7pPCaAqO46MgnOM80zW1RWuH61DGLwZJEdK2Kadq2F9CUG65" crossorigin="anonymous">
+
+</head>
+
+<body>
+
+    <p class="title">
+        EvoSeed: Unveiling the Threat on Deep Neural Networks with Real-World Illusions
+    </p>
+
+    <div style="text-align: center; font-size: 40pt; margin-bottom: 30px">
+            <!-- <span>CVPR 2023</span> -->
+    </div>
+
+    <p class="author">
+        <span class="author"><a target="_blank" href="https://sites.google.com/site/shashankkotyan">Shashank&nbsp;Kotyan</a>&nbsp;<sup>*</sup></span>
+        <span class="author">PoYuan&nbsp;Mao&nbsp;<sup>*</sup></span>
+        <span class="author"><a target="_blank" href="http://danilovargas.org/">Danilo&nbsp;Vasconcellos&nbsp;Vargas</a></span>
+    </p>
+
+    <table class="affiliations">
+        <tbody>
+            <tr>
+                <td style="text-align: center; width:0%; ">Kyushu University</td>
+            </tr>
+            <tr>
+                <td style="text-align: center; width:0%; "><sup>*</sup>&nbsp;Equal Contribution</td>
+            </tr>
+        </tbody>
+    </table>
+
+    <table align="center" class="menu">
+        <tbody>
+            <tr>
+                <td align="center"> 
+                    <span class="menu"><a href="" target="_blank">[Paper]</a></span>
+                    <span class="menu"><a href="https://github.com/shashankkotyan/EvoSeed/">[Code]</a></span>
+                    <span class="menu"><a href="./assets/reference.bib">[BibTeX]</a></span>
+                </td> 
+            </tr>
+        </tbody>
+    </table>
+
+    <div class="container">
+        <br><hr class="hr-twill-colorful"><br>
+
+        <div class="image" style="text-align:center;">
+            <img src="./assets/illustration.png" alt="Overview of three distinct basic patterns of k* Distribution." width="60%" style="margin: auto" />
+            <figcaption>Figure: Performance of EvoSeed compared to Random Search.</figcaption>
+        </div>
+
+        <br>
+
+        <strong>Key Contributions:</strong>
+
+        <ul>
+            <li> A model-agnostic black-box algorithimic framework based on Evolutionary Strategy to generate unrestricted natural adversarial samples. </li>
+            <li> High-Quality Natural Adversarial Samples that also show potential to misclassify variety of robust classifiers. </li>
+        </ul>
+
+        <br><hr class="hr-twill-colorful"><br>
+
+        <p class="section"><strong>Abstract</strong></p> 
+        <p class="text"> 
+            Deep neural networks are exploited using natural adversarial samples, which have no impact on human perception but are misclassified.
+            Current approaches often rely on the white-box nature of deep neural networks to generate these adversarial samples or alter the distribution of adversarial samples compared to training distribution.
+            To alleviate the limitations of current approaches, we propose EvoSeed, a novel evolutionary strategy-based search algorithmic framework to generate natural adversarial samples.
+            Our EvoSeed framework uses auxiliary Diffusion and Classifier models to operate in a model-agnostic black-box setting.
+            We employ CMA-ES to optimize the search for an adversarial seed vector, which, when processed by the Conditional Diffusion Model, results in an unrestricted natural adversarial sample misclassified by the Classifier Model.
+            Experiments show that generated adversarial images are of high image quality and are transferable to different classifiers.
+            Our approach demonstrates promise in enhancing the quality of adversarial samples using evolutionary algorithms.
+            We hope our research opens new avenues to enhance the robustness of deep neural networks in real-world scenarios. 
+        </p>    
+
+        <br><hr class="hr-twill-colorful"><br>
+
+        <p class="section"><strong>Method</strong></p>
+
+        <div class="image">
+            <img src="./assets/framework.png" alt="" width="100%"/>
+            <figcaption class="caption">Figure: Overview of the Evoseed framework to generate Natural Adversarial Samples. We use auxillary diffusion and classifier models in our framework.</figcaption>
+        </div>
+        <br>
+
+        <br><hr class="hr-twill-colorful"><br>
+
+        <p class="section"><strong>Visualizations of Generated Natural Adversarial Samples </strong></p>
+
+        <div class="image">
+            <img src="./assets/compare.png" alt="" width="100%">
+            <figcaption> Figure: Visualization of the distribution of samples in latent space using, k* distribution, and Dimensionality Reduction techniques like t-SNE, Isomap, PCA, and UMAP of all classes of 16-class-ImageNet for the Logit Layer of ResNet-50 Architecture.</figcaption> 
+        </div>
+
+        <br><hr class="hr-twill-colorful"><br>
+
+        <p class="section" id="bibtex"><b>Bibtex</b></p>
+        <pre class="bibtex">
+@article{kotyan2024evoseed,
+    title = {EvoSeed: Unveiling the Threat on Deep Neural Networks with Real-World Illusions,
+    shorttitle = {EvoSeed},
+    author = {Kotyan, Shashank and Mao, PoYuan and Vargas, Danilo Vasconcellos},
+    year = {2024},
+    month = feb,
+}
+        </pre>
+
+    </div>
+
+
+
+</body>
+
+<script src="https://ajax.googleapis.com/ajax/libs/jquery/3.4.1/jquery.min.js"></script>
+<script src="https://code.jquery.com/jquery-3.2.1.slim.min.js" integrity="sha384-KJ3o2DKtIkvYIK3UENzmM7KCkRr/rE9/Qpg6aAZGJwFDMVNA/GpGFF93hXpG5KkN" crossorigin="anonymous"></script>
+<script src="https://cdn.jsdelivr.net/npm/@popperjs/core@2.11.6/dist/umd/popper.min.js" integrity="sha384-oBqDVmMz9ATKxIep9tiCxS/Z9fNfEXiDAYTujMAeBAsjFuCZSmKbSSUnQlmh/jp3" crossorigin="anonymous"></script>
+<script src="https://cdn.jsdelivr.net/npm/bootstrap@5.2.3/dist/js/bootstrap.bundle.min.js" integrity="sha384-kenU1KFdBIe4zVF0s0G1M5b4hcpxyD9F7jL+jjXkk+Q2h455rYXK/7HAuoJl+0I4" crossorigin="anonymous"></script>
+</html>