You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

perturb_config.py 3.3 kB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546
  1. # Copyright 2021 Huawei Technologies Co., Ltd
  2. #
  3. # Licensed under the Apache License, Version 2.0 (the "License");
  4. # you may not use this file except in compliance with the License.
  5. # You may obtain a copy of the License at
  6. #
  7. # http://www.apache.org/licenses/LICENSE-2.0
  8. #
  9. # Unless required by applicable law or agreed to in writing, software
  10. # distributed under the License is distributed on an "AS IS" BASIS,
  11. # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  12. # See the License for the specific language governing permissions and
  13. # limitations under the License.
  14. """
  15. Configuration of natural robustness methods for server.
  16. """
  17. PerturbConfig = [{"method": "Contrast", "params": {"alpha": 1.5, "beta": 0}},
  18. {"method": "GaussianBlur", "params": {"ksize": 5}},
  19. {"method": "SaltAndPepperNoise", "params": {"factor": 0.05}},
  20. {"method": "Translate", "params": {"x_bias": 0.1, "y_bias": -0.2}},
  21. {"method": "Scale", "params": {"factor_x": 0.7, "factor_y": 0.7}},
  22. {"method": "Shear", "params": {"factor": 2, "direction": "horizontal"}},
  23. {"method": "Rotate", "params": {"angle": 40}},
  24. {"method": "MotionBlur", "params": {"degree": 5, "angle": 45}},
  25. {"method": "GradientBlur", "params": {"point": [50, 100], "kernel_num": 3, "center": True}},
  26. {"method": "GradientLuminance", "params": {"color_start": [255, 255, 255], "color_end": [0, 0, 0],
  27. "start_point": [100, 150], "scope": 0.3,
  28. "bright_rate": 0.3, "pattern": "light", "mode": "circle"}},
  29. {"method": "GradientLuminance", "params": {"color_start": [255, 255, 255],
  30. "color_end": [0, 0, 0], "start_point": [150, 200],
  31. "scope": 0.3, "pattern": "light", "mode": "horizontal"}},
  32. {"method": "GradientLuminance", "params": {"color_start": [255, 255, 255], "color_end": [0, 0, 0],
  33. "start_point": [150, 200], "scope": 0.3,
  34. "pattern": "light", "mode": "vertical"}},
  35. {"method": "Perlin", "params": {"ratio": 0.5, "shade": 0.1}},
  36. {"method": "Curve", "params": {"curves": 10, "depth": 10, "mode": "vertical"}},
  37. {"method": "BackgroundWord", "params": {"shade": 0.1}},
  38. {"method": "Perspective", "params": {"ori_pos": [[0, 0], [0, 800], [800, 0], [800, 800]],
  39. "dst_pos": [[50, 0], [0, 800], [780, 0], [800, 800]]}},
  40. {"method": "BackShadow", "params": {"back_type": 'leaf', "shade": 0.2}},
  41. {"method": "BackShadow", "params": {"back_type": 'window', "shade": 0.2}},
  42. {"method": "BackShadow", "params": {"back_type": 'person', "shade": 0.1}},
  43. {"method": "BackShadow", "params": {"back_type": 'background', "shade": 0.1}},
  44. ]

MindArmour关注AI的安全和隐私问题。致力于增强模型的安全可信、保护用户的数据隐私。主要包含3个模块:对抗样本鲁棒性模块、Fuzz Testing模块、隐私保护与评估模块。 对抗样本鲁棒性模块 对抗样本鲁棒性模块用于评估模型对于对抗样本的鲁棒性,并提供模型增强方法用于增强模型抗对抗样本攻击的能力,提升模型鲁棒性。对抗样本鲁棒性模块包含了4个子模块:对抗样本的生成、对抗样本的检测、模型防御、攻防评估。