ó ¾÷Xc@@sÐddlmZddlmZddlmZddlmZddlmZddlmZddl m Z dd l m Z dd l m Z dd lmZd e fd „ƒYZde fd„ƒYZdS(i(tabsolute_importi(tbackend(t activations(t initializers(t regularizers(t constraints(tLayer(t InputSpec(t conv_utils(t interfacestLocallyConnected1DcB@sheZdZejddd d eddd d d d d d„ ƒZd„Zd„Z d„Z d „Z RS( s³ Locally-connected layer for 1D inputs. The `LocallyConnected1D` layer works similarly to the `Conv1D` layer, except that weights are unshared, that is, a different set of filters is applied at each different patch of the input. # Example ```python # apply a unshared weight convolution 1d of length 3 to a sequence with # 10 timesteps, with 64 output filters model = Sequential() model.add(LocallyConnected1D(64, 3, input_shape=(10, 32))) # now model.output_shape == (None, 8, 64) # add a new conv1d on top model.add(LocallyConnected1D(32, 3)) # now model.output_shape == (None, 6, 32) ``` # Arguments filters: Integer, the dimensionality of the output space (i.e. the number output of filters in the convolution). kernel_size: An integer or tuple/list of a single integer, specifying the length of the 1D convolution window. strides: An integer or tuple/list of a single integer, specifying the stride length of the convolution. Specifying any stride value != 1 is incompatible with specifying any `dilation_rate` value != 1. padding: One of `"valid"` or `"same"` (case-insensitive). activation: Activation function to use (see [activations](../activations.md)). If you don't specify anything, no activation is applied (ie. "linear" activation: `a(x) = x`). use_bias: Boolean, whether the layer uses a bias vector. kernel_initializer: Initializer for the `kernel` weights matrix (see [initializers](../initializers.md)). bias_initializer: Initializer for the bias vector (see [initializers](../initializers.md)). kernel_regularizer: Regularizer function applied to the `kernel` weights matrix (see [regularizer](../regularizers.md)). bias_regularizer: Regularizer function applied to the bias vector (see [regularizer](../regularizers.md)). activity_regularizer: Regularizer function applied to the output of the layer (its "activation"). (see [regularizer](../regularizers.md)). kernel_constraint: Constraint function applied to the kernel matrix (see [constraints](../constraints.md)). bias_constraint: Constraint function applied to the bias vector (see [constraints](../constraints.md)). # Input shape 3D tensor with shape: `(batch_size, steps, input_dim)` # Output shape 3D tensor with shape: `(batch_size, new_steps, filters)` `steps` value might have changed due to padding or strides. itvalidtglorot_uniformtzeroscK@sDtt|ƒj|||_tj|ddƒ|_tj|ddƒ|_tj|ƒ|_ |j dkrƒt d|ƒ‚ntj |ƒ|_ t j|ƒ|_||_tj|ƒ|_tj| ƒ|_tj| ƒ|_tj| ƒ|_tj| ƒ|_tj| ƒ|_tj|ƒ|_tddƒ|_dS(Nit kernel_sizetstridesR sHInvalid border mode for LocallyConnected1D (only "valid" is supported): tndimi(tsuperR t__init__tfiltersRtnormalize_tupleRRtnormalize_paddingtpaddingt ValueErrortnormalize_data_formatt data_formatRtgett activationtuse_biasRtkernel_initializertbias_initializerRtkernel_regularizertbias_regularizertactivity_regularizerRtkernel_constrainttbias_constraintRt input_spec(tselfRRRRRRRRRRR R!R"R#tkwargs((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyRKs&   c C@s(|d}|dkr(td|ƒ‚ntj|d|jd|j|jdƒ}||jd||jf|_|j |jd|j ddd|j d |j ƒ|_ |jró|j ||jfd|jdd d|jd |jƒ|_n d|_td d d i|d6ƒ|_t|_dS(Nis5Axis 2 of input should be fully-defined. Found shape:iit initializertnametkernelt regularizert constrainttbiasRitaxes(tNoneRRtconv_output_lengthRRRRt kernel_shapet add_weightRRR"R)RRR R#R,RR$tTruetbuilt(R%t input_shapet input_dimt output_length((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pytbuildos6            cC@sAtj|d|jd|j|jdƒ}|d||jfS(Nii(RR/RRRR(R%R4tlength((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pytcompute_output_shapes   c C@s(|jd}|j\}}}g}xqt|ƒD]c}t|||||jdƒ}|jtj|dd…|dd…fdd|fƒƒq2Wtj|ddƒ} tj | |j ƒ} tj | dƒ} |j r| tj|j d||fƒ7} n|jdk r$|j| ƒ} n| S(Niiiÿÿÿÿtaxisi(iii(RR0trangetsliceRtappendtKtreshapet concatenatet batch_dotR)tpermute_dimensionsRR,RR.( R%tinputststrideR6t feature_dimRtxstit slice_lengtht x_aggregatetoutput((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pytcall”s   ( %cC@s i |jd6|jd6|jd6|jd6tj|jƒd6|jd6tj|j ƒd6tj|j ƒd6t j|j ƒd 6t j|j ƒd 6t j|jƒd 6tj|jƒd 6tj|jƒd 6}tt|ƒjƒ}tt|jƒƒt|jƒƒƒS(NRRRRRRRRRR R!R"R#(RRRRRt serializeRRRRRRRR R!RR"R#RR t get_configtdicttlisttitems(R%tconfigt base_config((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyRM©s      N( t__name__t __module__t__doc__R tlegacy_conv1d_supportR.R2RR7R9RKRM(((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyR s$:   tLocallyConnected2DcB@sheZdZejd dd d eddd d d d d d„ ƒZd„Zd„Z d„Z d „Z RS( s­Locally-connected layer for 2D inputs. The `LocallyConnected2D` layer works similarly to the `Conv2D` layer, except that weights are unshared, that is, a different set of filters is applied at each different patch of the input. # Examples ```python # apply a 3x3 unshared weights convolution with 64 output filters on a 32x32 image # with `data_format="channels_last"`: model = Sequential() model.add(LocallyConnected2D(64, (3, 3), input_shape=(32, 32, 3))) # now model.output_shape == (None, 30, 30, 64) # notice that this layer will consume (30*30)*(3*3*3*64) + (30*30)*64 parameters # add a 3x3 unshared weights convolution on top, with 32 output filters: model.add(LocallyConnected2D(32, (3, 3))) # now model.output_shape == (None, 28, 28, 32) ``` # Arguments filters: Integer, the dimensionality of the output space (i.e. the number output of filters in the convolution). kernel_size: An integer or tuple/list of 2 integers, specifying the width and height of the 2D convolution window. Can be a single integer to specify the same value for all spatial dimensions. strides: An integer or tuple/list of 2 integers, specifying the strides of the convolution along the width and height. Can be a single integer to specify the same value for all spatial dimensions. Specifying any stride value != 1 is incompatible with specifying any `dilation_rate` value != 1. padding: one of `"valid"` or `"same"` (case-insensitive). data_format: A string, one of `channels_last` (default) or `channels_first`. The ordering of the dimensions in the inputs. `channels_last` corresponds to inputs with shape `(batch, height, width, channels)` while `channels_first` corresponds to inputs with shape `(batch, channels, height, width)`. It defaults to the `image_data_format` value found in your Keras config file at `~/.keras/keras.json`. If you never set it, then it will be "channels_last". activation: Activation function to use (see [activations](../activations.md)). If you don't specify anything, no activation is applied (ie. "linear" activation: `a(x) = x`). use_bias: Boolean, whether the layer uses a bias vector. kernel_initializer: Initializer for the `kernel` weights matrix (see [initializers](../initializers.md)). bias_initializer: Initializer for the bias vector (see [initializers](../initializers.md)). kernel_regularizer: Regularizer function applied to the `kernel` weights matrix (see [regularizer](../regularizers.md)). bias_regularizer: Regularizer function applied to the bias vector (see [regularizer](../regularizers.md)). activity_regularizer: Regularizer function applied to the output of the layer (its "activation"). (see [regularizer](../regularizers.md)). kernel_constraint: Constraint function applied to the kernel matrix (see [constraints](../constraints.md)). bias_constraint: Constraint function applied to the bias vector (see [constraints](../constraints.md)). # Input shape 4D tensor with shape: `(samples, channels, rows, cols)` if data_format='channels_first' or 4D tensor with shape: `(samples, rows, cols, channels)` if data_format='channels_last'. # Output shape 4D tensor with shape: `(samples, filters, new_rows, new_cols)` if data_format='channels_first' or 4D tensor with shape: `(samples, new_rows, new_cols, filters)` if data_format='channels_last'. `rows` and `cols` values might have changed due to padding. iR R R cK@sDtt|ƒj|||_tj|ddƒ|_tj|ddƒ|_tj|ƒ|_ |j dkrƒt d|ƒ‚ntj |ƒ|_ t j|ƒ|_||_tj|ƒ|_tj| ƒ|_tj| ƒ|_tj| ƒ|_tj| ƒ|_tj| ƒ|_tj|ƒ|_tddƒ|_dS(NiRRR sHInvalid border mode for LocallyConnected2D (only "valid" is supported): Ri(RRWRRRRRRRRRRRRRRRRRRRRR R!RR"R#RR$(R%RRRRRRRRRRR R!R"R#R&((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyRs&   c C@sô|jdkr/|dd!\}}|d}n|d\}}|d}|dksa|dkrztdt|ƒƒ‚ntj||jd|j|jdƒ}tj||jd|j|jdƒ}||_ ||_ |||jd|jd||j f|_ |j |j d|jd d d |jd |jƒ|_|jrŽ|j |||j fd|jd d d |jd |jƒ|_n d|_|jdkrÈtdddi|d6ƒ|_ntdddi|d6ƒ|_t|_dS(Nt channels_lastiiÿÿÿÿiisThe spatial dimensions of the inputs to a LocallyConnected2D layer should be fully-defined, but layer received the inputs shape iR'R(R)R*R+R,tchannels_firstRiR-(RR.RtstrRR/RRRt output_rowt output_colRR0R1RRR"R)RRR R#R,RR$R2R3(R%R4t input_rowt input_colt input_filterR[R\((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyR73sB          "cC@sî|jdkr&|d}|d}n&|jdkrL|d}|d}ntj||jd|j|jdƒ}tj||jd|j|jdƒ}|jdkrÄ|d|j||fS|jdkrê|d|||jfSdS(NRYiiRXii(RRR/RRRR(R%R4trowstcols((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyR9[s    c C@s5|j\}}|j\}}}|jdkrntjƒdkrOg}xìt|jƒD]Û}xÒt|jƒD]Á} t|||||j dƒ} t| || ||j dƒ} tj |dd…dd…| | fdd|fƒ} |j tj | |j ||j| dd…dd…fƒƒqnWqXWtj|ddƒ}næg} x³t|jƒD]¢}x™t|jƒD]ˆ} t|||||j dƒ} t| || ||j dƒ} | j tj |dd…dd…| | fdd|fƒƒq{WqeWtj| ddƒ}tj||j ƒ}tj ||j|jd|fƒ}tj|d ƒ}n.|jd krœg} x³t|jƒD]¢}x™t|jƒD]ˆ} t|||||j dƒ} t| || ||j dƒ} | j tj |dd…| | dd…fdd|fƒƒq©Wq“Wtj| ddƒ}tj||j ƒ}tj ||j|jd|fƒ}tj|d ƒ}n|jr"|jdkrâ|tj |jd||j|jfƒ7}q"|jd kr"|tj |jd|j|j|fƒ7}q"n|j|ƒ}|S( NRYttheanoiiiÿÿÿÿR:iiRX(iiii(iiii(RR0RR>RR;R[R\R<RR?R=tdotR)R@RARBRR,R(R%RCt stride_rowt stride_colt_RERRJRGtjt slice_rowt slice_colt x_flattenRFRI((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyRKmsl  %9  +   +  "cC@si|jd6|jd6|jd6|jd6|jd6tj|jƒd6|jd6t j|j ƒd6t j|j ƒd 6t j|j ƒd 6t j|jƒd 6t j|jƒd 6tj|jƒd 6tj|jƒd6}tt|ƒjƒ}tt|jƒƒt|jƒƒƒS(NRRRRRRRRRRR R!R"R#(RRRRRRRLRRRRRRRR R!RR"R#RRWRMRNRORP(R%RQRR((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyRM©s"      (iiN( RSRTRUR tlegacy_conv2d_supportR.R2RR7R9RKRM(((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyRW½s$P (  RRRRtengineRRtutilsRtlegacyR R RW(((s1/tmp/pip-build-isqEY4/keras/keras/layers/local.pyts®