File size: 10,888 Bytes
ccf396a
 
 
 
 
a767b82
 
ccf396a
a767b82
ccf396a
 
 
 
 
 
 
 
 
 
 
a767b82
 
 
 
ccf396a
 
 
 
 
 
 
 
 
 
 
 
 
 
a767b82
 
ccf396a
 
a767b82
ccf396a
a767b82
ccf396a
a767b82
ccf396a
 
a767b82
ccf396a
 
a767b82
ccf396a
 
 
 
a767b82
 
 
 
ccf396a
 
 
 
 
a767b82
ccf396a
 
 
 
 
 
 
 
 
a767b82
ccf396a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
a767b82
 
 
ccf396a
a767b82
ccf396a
 
 
 
 
 
 
 
 
 
 
 
 
 
a767b82
ccf396a
 
a767b82
ccf396a
 
 
 
 
 
 
 
 
 
 
 
 
a767b82
 
 
 
ccf396a
 
a767b82
 
 
 
ccf396a
a767b82
 
 
ccf396a
 
a767b82
 
 
 
ccf396a
 
a767b82
 
 
ccf396a
 
a767b82
 
ccf396a
 
a767b82
 
ccf396a
 
a767b82
 
ccf396a
 
 
a767b82
 
 
 
 
ccf396a
 
 
 
 
 
 
a767b82
 
 
 
 
 
 
ccf396a
 
 
 
 
 
 
 
a767b82
ccf396a
 
a767b82
ccf396a
 
 
 
 
 
 
a767b82
ccf396a
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
<!DOCTYPE html>
<html>
<head>
  <meta charset="utf-8">
  <meta name="description"
        content="NeuralFuse provides model-independent protection for AI accelerators built on a chip, allowing them to maintain stable performance when suffering low-voltage-induced bit errors.">
  <meta name="keywords" content="machine learning, energy efficient inference, bit error resilience">
  <meta name="viewport" content="width=device-width, initial-scale=1">
  <title>NeuralFuse: Learning to Recover the Accuracy of Access-Limited Neural Network Inference in Low-Voltage Regimes</title>

  <link href="https://fonts.googleapis.com/css?family=Google+Sans|Noto+Sans|Castoro"
        rel="stylesheet">

  <link rel="stylesheet" href="./static/css/bulma.min.css">
  <link rel="stylesheet" href="./static/css/bulma-carousel.min.css">
  <link rel="stylesheet" href="./static/css/bulma-slider.min.css">
  <link rel="stylesheet" href="./static/css/fontawesome.all.min.css">
  <link rel="stylesheet"
        href="https://cdn.jsdelivr.net/gh/jpswalsh/academicons@1/css/academicons.min.css">
  <link rel="stylesheet" href="./static/css/index.css">
  <link rel="apple-touch-icon" sizes="180x180" href="./static/images/favicon/apple-touch-icon.png">
  <link rel="icon" type="image/png" sizes="32x32" href="./static/images/favicon/favicon-32x32.png">
  <link rel="icon" type="image/png" sizes="16x16" href="./static/images/favicon/favicon-16x16.png">
  <link rel="manifest" href="./static/images/favicon/site.webmanifest">

  <script src="https://ajax.googleapis.com/ajax/libs/jquery/3.5.1/jquery.min.js"></script>
  <script defer src="./static/js/fontawesome.all.min.js"></script>
  <script src="./static/js/bulma-carousel.min.js"></script>
  <script src="./static/js/bulma-slider.min.js"></script>
  <script src="./static/js/index.js"></script>
</head>
<body>

<section class="hero">
  <div class="hero-body">
    <div class="container is-max-desktop">
      <div class="columns is-centered">
        <div class="column has-text-centered">
          <h1 class="title is-1 publication-title">✨NeuralFuse✨</h1>
          <h1 class="title publication-subtitle">Learning to Recover the Accuracy of Access-Limited Neural Network Inference in Low-Voltage Regimes</h1>
          <div class="is-size-5 publication-authors">
            <span class="author-block">
              <a href="https://scholar.google.com/citations?user=g2MolmMAAAAJ&hl=en" target="_blank">Hao-Lun Sun</a><sup>1</sup>,</span>
            <span class="author-block">
              <a href="https://hsiung.cc" target="_blank">Lei Hsiung</a><sup>2</sup>,</span>
            <span class="author-block">
              <a href="https://scholar.google.com/citations?user=qurg568AAAAJ&hl=en" target="_blank">Nandhini Chandramoorthy</a><sup>3</sup>,
            </span>
            <span class="author-block">
              <a href="https://sites.google.com/site/pinyuchenpage/home" target="_blank">Pin-Yu Chen</a><sup>3</sup>,
            </span>
            <span class="author-block">
              <a href="https://tsungyiho.github.io" target="_blank">Tsung-Yi Ho</a><sup>4</sup>,
            </span>
          </div>

          <div class="is-size-5 publication-authors">
            <span class="author-block"><sup>1</sup>National Tsing Hua University</span>
            <span class="author-block"><sup>2</sup>Dartmouth College</span>
            <span class="author-block"><sup>3</sup>IBM Research</span>
            <span class="author-block"><sup>4</sup>CUHK</span>
          </div>

          <div class="column has-text-centered">
            <div class="publication-links">
              <span class="link-block">
                <a href="https://arxiv.org/abs/2306.16869" target="_blank"
                   class="external-link button is-normal is-rounded is-dark">
                  <span class="icon">
                      <i class="ai ai-arxiv"></i>
                  </span>
                  <span>arXiv</span>
                </a>
              </span>
              <!-- Code Link. -->
              <span class="link-block">
                <a href="https://github.com/IBM/NeuralFuse" target="_blank"
                   class="external-link button is-normal is-rounded is-dark">
                  <span class="icon">
                      <i class="fab fa-github"></i>
                  </span>
                  <span>Code</span>
                  </a>
              </span>
            </div>

          </div>
        </div>
      </div>
    </div>
  </div>
</section>

<section class="hero teaser">
  <div class="container is-max-desktop">
    <div class="hero-body">
      <img id="teaser" src="./static/images/teaser.png"
           class=""
           alt="NeuralFuse Teaser"/>
      <h2 class="subtitle has-text-centered">
        The pipeline of the <span class="small_caps">NeuralFuse</span> framework at inference.
      </h2>
    </div>
  </div>
</section>


<section class="section">
  <div class="container is-max-desktop">
    <!-- Abstract. -->
    <div class="columns is-centered has-text-centered">
      <div class="column is-four-fifths">
        <h2 class="title is-3">Abstract</h2>
        <div class="content has-text-justified">
          <p>
            Deep neural networks (DNNs) have become ubiquitous in machine learning, but their energy consumption remains problematically high. An effective strategy for reducing such consumption is supply-voltage reduction, but if done too aggressively, it can lead to accuracy degradation. This is due to random bit-flips in static random access memory (SRAM), where model parameters are stored.
          </p>
          <p>
            To address this challenge, we have developed <span class="small_caps">NeuralFuse</span>, a novel add-on module that handles the energy-accuracy tradeoff in low-voltage regimes by learning input transformations and using them to generate error-resistant data representations, thereby protecting DNN accuracy in both nominal and low-voltage scenarios. As well as being easy to implement, NeuralFuse can be readily applied to DNNs with limited access, such cloud-based APIs that are accessed remotely or non-configurable hardware. Our experimental results demonstrate that, at a 1% bit-error rate, NeuralFuse can reduce SRAM access energy by up to 24% while recovering accuracy by up to 57%. To the best of our knowledge, this is the first approach to addressing low-voltage-induced bit errors that requires no model retraining.
          </p>
        </div>
      </div>
    </div>
    <!--/ Abstract. -->
  </div>
</section>


<section class="section">
  <div class="container is-max-desktop">
    <div class="columns is-centered">
      <div class="column is-full-width">
        <h2 class="title is-3">Our Contributions</h2>
        <div class="content has-text-justified highlight-box">
          <p><span class="contribution-subtitle">Boosts DNN Accuracy Under Low Power</span>
              <span class="small_caps">NeuralFuse</span> improves the accuracy of deep neural networks (DNNs) operating in low-power environments with random bit errors, without needing to retrain the models.
          </p>
        </div>
        <div class="content has-text-justified highlight-box">
          <p><span class="contribution-subtitle">Protects DNN Accuracy Under Unstable Power</span>
              <span class="small_caps">NeuralFuse</span> improves the accuracy of deep neural networks (DNNs) operating in low-power environments with random bit errors, without needing to retrain the models.
          </p>
        </div>
        <div class="content has-text-justified highlight-box">
          <p><span class="contribution-subtitle">Adapts to Limited-Access Settings</span>
               <span class="small_caps">NeuralFuse</span> supports deployment in scenarios with limited access to model details, using flexible training methods to adapt effectively across diverse DNN architectures.
          </p>
        </div>
        <div class="content has-text-justified highlight-box">
          <p><span class="contribution-subtitle">Reduces Energy Use with Proven Performance</span>
              <span class="small_caps">NeuralFuse</span> recovers up to 57% of lost accuracy and reduces memory access energy by up to 24%, tested across diverse models (ResNet18, ResNet50, VGG11, VGG16, and VGG19) and datasets (CIFAR-10, CIFAR-100, GTSRB, and ImageNet-10).
          </p>
        </div>
      </div>
    </div>    
  </div>
</section>


<section class="section">
  <div class="container is-max-desktop">
    <div class="columns is-centered">
      <div class="column is-full-width">
        <h2 class="title is-3">NeuralFuse Performance</h2>
        <h3 class="title is-4">Energy/Accuracy Tradeoff</h3>
        <div class="content has-text-justified">
          <p>
            On the same base model (ResNet18), we illustrate the energy/accuracy tradeoff of six NeuralFuse implementations.
The x-axis represents the percentage reduction in dynamic-memory access energy at low-voltage settings (base model protected by NeuralFuse), as compared to the bit-error-free (nominal) voltage. The y-axis represents the perturbed accuracy (evaluated at low voltage) with a 1% bit-error rate.
          </p>
        </div>

        <img id="performance" src="./static/images/performance.png"
             class=""
             alt="NeuralFuse Performance"/>
      </div>
    </div>    
  </div>
</section>


<section class="section" id="BibTeX">
  <div class="container is-max-desktop content">
    <h2 class="title">BibTeX</h2>
    <pre><code>@inproceedings{sun2024neuralfuse,
  title={{NeuralFuse: Learning to Recover the Accuracy of Access-Limited Neural Network Inference in Low-Voltage Regimes}},
  author={Hao-Lun Sun and Lei Hsiung and Nandhini Chandramoorthy and Pin-Yu Chen and Tsung-Yi Ho},
  booktitle = {Advances in Neural Information Processing Systems},
  publisher = {Curran Associates, Inc.},
  volume = {37},
  year = {2024}
}</code></pre>
  </div>
</section>


<footer class="footer">
  <div class="container">
    <div class="content has-text-centered">
      <a class="icon-link" href="https://arxiv.org/abs/2306.16869" target="_blank" class="external-link">
        <i class="fas fa-file-pdf"></i>
      </a>
      <a class="icon-link" href="https://github.com/IBM/NeuralFuse" target="_blank" class="external-link">
        <i class="fab fa-github"></i>
      </a>
    </div>
    <div class="columns is-centered">
      <div class="column is-8">
        <div class="content">
          <p>
            This page is maintained by <a target="_blank" href="https://hsiung.cc">Lei Hsiung</a>. Page template is borrowed from <a target="_blank" rel="nofollow" href="https://github.com/nerfies/nerfies.github.io">here</a>.
          </p>
        </div>
      </div>
    </div>
  </div>
</footer>

</body>
</html>