layers

Type Members

1. class Activation[T] extends bigdl.nn.keras.Activation[T] with Net

   Simple activation function to be applied to the output.

2. class AddConstant[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

   Add a (non-learnable) scalar constant to the input.

3. class AtrousConvolution1D[T] extends bigdl.nn.keras.AtrousConvolution1D[T] with Net

   Applies an atrous convolution operator for filtering neighborhoods of 1-D inputs.

4. class AtrousConvolution2D[T] extends bigdl.nn.keras.AtrousConvolution2D[T] with Net

   Applies an atrous convolution operator for filtering windows of 2-D inputs.

5. class AveragePooling1D[T] extends Pooling1D[T] with Net

   Applies average pooling operation for temporal data.

6. class AveragePooling2D[T] extends Pooling2D[T] with Net

   Applies average pooling operation for spatial data.

7. class AveragePooling3D[T] extends Pooling3D[T] with Net

   Applies average pooling operation for 3D data (spatial or spatio-temporal).

8. class BatchNormalization[T] extends bigdl.nn.keras.BatchNormalization[T] with Net

   Batch normalization layer.

9. class Bidirectional[T] extends bigdl.nn.keras.Bidirectional[T] with Net

   Bidirectional wrapper for RNNs.

10. class BinaryThreshold[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Threshold the input.

11. class CAdd[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    This layer has a bias with given size.

12. class CMul[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    This layer has a weight with given size.

13. class ConvLSTM2D[T] extends bigdl.nn.keras.ConvLSTM2D[T] with Net

    Convolutional LSTM.

14. class Convolution1D[T] extends bigdl.nn.keras.Convolution1D[T] with Net

    Applies convolution operator for filtering neighborhoods of 1-D inputs.

15. class Convolution2D[T] extends bigdl.nn.keras.Convolution2D[T] with Net

    Applies a 2D convolution over an input image composed of several input planes.

16. class Convolution3D[T] extends bigdl.nn.keras.Convolution3D[T] with Net

    Applies convolution operator for filtering windows of three-dimensional inputs.

17. class Cropping1D[T] extends bigdl.nn.keras.Cropping1D[T] with Net

    Cropping layer for 1D input (e.

18. class Cropping2D[T] extends bigdl.nn.keras.Cropping2D[T] with Net

    Cropping layer for 2D input (e.

19. class Cropping3D[T] extends bigdl.nn.keras.Cropping3D[T] with Net

    Cropping layer for 3D data (e.

20. class Deconvolution2D[T] extends bigdl.nn.keras.Deconvolution2D[T] with Net

    Transposed convolution operator for filtering windows of 2-D inputs.

21. class Dense[T] extends bigdl.nn.keras.Dense[T] with Net

    A densely-connected NN layer.

22. class Dropout[T] extends bigdl.nn.keras.Dropout[T] with Net

    Applies Dropout to the input by randomly setting a fraction 'p' of input units to 0 at each update during training time in order to prevent overfitting.

23. class ELU[T] extends bigdl.nn.keras.ELU[T] with Net

    Exponential Linear Unit.

24. class Embedding[T] extends bigdl.nn.keras.Embedding[T] with Net

    Turn positive integers (indexes) into dense vectors of fixed size.

25. class Exp[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies element-wise exp to the input.

26. class Flatten[T] extends bigdl.nn.keras.Flatten[T] with Net

    Flattens the input without affecting the batch size.

27. class GRU[T] extends Recurrent[T] with Net

    Gated Recurrent Unit architecture.

28. class GaussianDropout[T] extends bigdl.nn.keras.GaussianDropout[T] with Net

    Apply multiplicative 1-centered Gaussian noise.

29. class GaussianNoise[T] extends bigdl.nn.keras.GaussianNoise[T] with Net

    Apply additive zero-centered Gaussian noise.

30. class GaussianSampler[T] extends LayerWrapperByForward[T]

    Takes {mean, log_variance} as input and samples from the Gaussian distribution

31. class GlobalAveragePooling1D[T] extends GlobalPooling1D[T] with Net

    Applies global average pooling operation for temporal data.

32. class GlobalAveragePooling2D[T] extends GlobalPooling2D[T] with Net

    Applies global average pooling operation for spatial data.

33. class GlobalAveragePooling3D[T] extends GlobalPooling3D[T] with Net

    Applies global average pooling operation for 3D data.

34. class GlobalMaxPooling1D[T] extends GlobalPooling1D[T] with Net

    Applies global max pooling operation for temporal data.

35. class GlobalMaxPooling2D[T] extends GlobalPooling2D[T] with Net

    Applies global max pooling operation for spatial data.

36. class GlobalMaxPooling3D[T] extends GlobalPooling3D[T] with Net

    Applies global max pooling operation for 3D data.

37. class HardShrink[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies the hard shrinkage function element-wise to the input.

38. class HardTanh[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies the hard tanh function element-wise to the input.

39. class Highway[T] extends bigdl.nn.keras.Highway[T] with Net

    Densely connected highway network.

40. class Input[T] extends KerasLayer[Activity, Activity, T] with Net

41. class KerasLayerWrapper[T] extends LayerWrapperByForward[T]

    Wrap a torch style layer to keras style layer.

42. class LRN2D[T] extends KerasLayer[Tensor[T], Tensor[T], T] with Net

    Local Response Normalization between different feature maps.

43. class LSTM[T] extends Recurrent[T] with Net

    Long Short Term Memory unit architecture.

44. abstract class LayerWrapperByForward[T] extends KerasLayer[Activity, Activity, T] with Net

45. class LeakyReLU[T] extends bigdl.nn.keras.LeakyReLU[T] with Net

    Leaky version of a Rectified Linear Unit.

46. class LocallyConnected1D[T] extends bigdl.nn.keras.LocallyConnected1D[T] with Net

    Locally-connected layer for 1D inputs which works similarly to the TemporalConvolution layer, except that weights are unshared, that is, a different set of filters is applied at each different patch of the input.

47. class LocallyConnected2D[T] extends bigdl.nn.keras.LocallyConnected2D[T] with Net

    Locally-connected layer for 2D inputs that works similarly to the SpatialConvolution layer, except that weights are unshared, that is, a different set of filters is applied at each different patch of the input.

48. class Log[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies a log transformation to the input.

49. class Masking[T] extends bigdl.nn.keras.Masking[T] with Net

    Use a mask value to skip timesteps for a sequence.

50. class MaxPooling1D[T] extends Pooling1D[T] with Net

    Applies max pooling operation for temporal data.

51. class MaxPooling2D[T] extends Pooling2D[T] with Net

    Applies max pooling operation for spatial data.

52. class MaxPooling3D[T] extends Pooling3D[T] with Net

    Applies max pooling operation for 3D data (spatial or spatio-temporal).

53. class MaxoutDense[T] extends bigdl.nn.keras.MaxoutDense[T] with Net

    A dense maxout layer that takes the element-wise maximum of linear layers.

54. class Merge[T] extends KerasLayer[Tensor[T], Tensor[T], T] with Net

    Used to merge a list of inputs into a single output, following some merge mode.

55. class Mul[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Multiply a single scalar factor to the input.

56. class MulConstant[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Multiply the input by a (non-learnable) scalar constant.

57. class Narrow[T] extends KerasLayer[Tensor[T], Tensor[T], T] with Net

    Narrow the input with the number of dimensions not being reduced.

58. class Negative[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Computes the negative value of each element of the input.

59. class PReLU[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies parametric ReLU, where parameter varies the slope of the negative part.

60. class Permute[T] extends bigdl.nn.keras.Permute[T] with Net

    Permutes the dimensions of the input according to a given pattern.

61. class Power[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies an element-wise power operation with scale and shift to the input.

62. class RReLU[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies the randomized leaky rectified linear unit element-wise to the input.

63. class RepeatVector[T] extends bigdl.nn.keras.RepeatVector[T] with Net

    Repeats the input n times.

64. class Reshape[T] extends KerasLayer[Tensor[T], Tensor[T], T]

    Reshapes an output to a certain shape.

65. class ResizeBilinear[T] extends LayerWrapperByForward[T]

    Resize the input image with bilinear interpolation.

66. class SReLU[T] extends bigdl.nn.keras.SReLU[T] with Net

    S-shaped Rectified Linear Unit.

67. class Scale[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape

    Scale is the combination of CMul and CAdd.

68. class Select[T] extends KerasLayer[Tensor[T], Tensor[T], T] with Net

    Select an index of the input in the given dim and return the subset part.

69. class SeparableConvolution2D[T] extends bigdl.nn.keras.SeparableConvolution2D[T] with Net

    Applies separable convolution operator for 2D inputs.

70. class ShareConvolution2D[T] extends KerasLayer[Tensor[T], Tensor[T], T] with Net

    Applies a 2D convolution over an input image composed of several input planes.

71. class SimpleRNN[T] extends Recurrent[T] with Net

    A fully-connected recurrent neural network cell.

72. class SoftShrink[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies the soft shrinkage function element-wise to the input.

73. class SpatialDropout1D[T] extends bigdl.nn.keras.SpatialDropout1D[T] with Net

    Spatial 1D version of Dropout.

74. class SpatialDropout2D[T] extends bigdl.nn.keras.SpatialDropout2D[T] with Net

    Spatial 2D version of Dropout.

75. class SpatialDropout3D[T] extends bigdl.nn.keras.SpatialDropout3D[T] with Net

    Spatial 3D version of Dropout.

76. class Sqrt[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies an element-wise square root operation to the input.

77. class Square[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Applies an element-wise square operation to the input.

78. class Squeeze[T] extends KerasLayer[Tensor[T], Tensor[T], T] with Net

    Delete the singleton dimension(s).

79. class Threshold[T] extends KerasLayer[Tensor[T], Tensor[T], T] with IdentityOutputShape with Net

    Threshold input Tensor.

80. class ThresholdedReLU[T] extends bigdl.nn.keras.ThresholdedReLU[T] with Net

    Thresholded Rectified Linear Unit.

81. class TimeDistributed[T] extends KerasLayer[Tensor[T], Tensor[T], T] with Net

    TimeDistributed wrapper.

82. class UpSampling1D[T] extends bigdl.nn.keras.UpSampling1D[T] with Net

    UpSampling layer for 1D inputs.

83. class UpSampling2D[T] extends bigdl.nn.keras.UpSampling2D[T] with Net

    UpSampling layer for 2D inputs.

84. class UpSampling3D[T] extends bigdl.nn.keras.UpSampling3D[T] with Net

    UpSampling layer for 3D inputs.

85. class WithinChannelLRN2D[T] extends KerasLayer[Tensor[T], Tensor[T], T] with Net

    The local response normalization layer performs a kind of "lateral inhibition" by normalizing over local input regions.

86. class ZeroPadding1D[T] extends bigdl.nn.keras.ZeroPadding1D[T] with Net

    Zero-padding layer for 1D input (e.

87. class ZeroPadding2D[T] extends bigdl.nn.keras.ZeroPadding2D[T] with Net

    Zero-padding layer for 2D input (e.

88. class ZeroPadding3D[T] extends bigdl.nn.keras.ZeroPadding3D[T] with Net

    Zero-padding layer for 3D data (spatial or spatio-temporal).