Update inputs via updates at scattered (sparse) indices.

At a high level, this operation does inputs[indices] <- updates. Assume inputs is a tensor of shape (D1, D2, ..., Dn), there are 2 main usages of scatter_update.

1. indices is a 2D tensor of shape (num_updates, n), where num_updates is the number of updates to perform, and updates is a 1D tensor of shape (num_updates). For example, if inputs is op_zeros(c(4, 4, 4)), and we want to update inputs[2, 3, 4] and inputs[1, 2, 4] as 1, then we can use:

inputs <- op_zeros(c(4, 4, 4))
indices <- rbind(c(2, 3, 4),
                 c(1, 2, 4))
updates <- op_array(c(1, 1), "float32")
op_scatter_update(inputs, indices, updates)

## tf.Tensor(
## [[[0. 0. 0. 0.]
##   [0. 0. 0. 1.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]]
##
##  [[0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 1.]
##   [0. 0. 0. 0.]]
##
##  [[0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]]
##
##  [[0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32)

2 indices is a 2D tensor of shape (num_updates, k), where num_updates is the number of updates to perform, and k (k <= n) is the size of each index in indices. updates is a n - k-D tensor of shape (num_updates, shape(inputs)[-(1:k)]). For example, if inputs <- op_zeros(c(4, 4, 4)), and we want to update inputs[1, 2, ] and inputs[2, 3, ] as [1, 1, 1, 1], then indices would have shape (num_updates, 2) (k = 2), and updates would have shape (num_updates, 4) (shape(inputs)[3:4] == 4). See the code below:

inputs <- op_zeros(c(4, 4, 4))
indices <- rbind(c(2, 3),
                 c(3, 4))
updates <- op_array(rbind(c(1, 1, 1, 1),
                          c(1, 1, 1, 1)),
                    "float32")
op_scatter_update(inputs, indices, updates)

## tf.Tensor(
## [[[0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]]
##
##  [[0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [1. 1. 1. 1.]
##   [0. 0. 0. 0.]]
##
##  [[0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [1. 1. 1. 1.]]
##
##  [[0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]
##   [0. 0. 0. 0.]]], shape=(4, 4, 4), dtype=float32)

Usage

op_scatter_update(inputs, indices, updates)

Arguments

inputs

A tensor, the tensor to be updated.

indices

A tensor or list of shape (N, inputs$ndim), specifying indices to update. N is the number of indices to update, must be equal to the first dimension of updates.

updates

A tensor, the new values to be put to inputs at indices.

Value

A tensor, has the same shape and dtype as inputs.

See also

• https://keras.io/api/ops/core#scatterupdate-function

Other core ops:
op_associative_scan()
op_cast()
op_cond()
op_convert_to_numpy()
op_convert_to_tensor()
op_custom_gradient()
op_dtype()
op_fori_loop()
op_is_tensor()
op_map()
op_scan()
op_scatter()
op_searchsorted()
op_shape()
op_slice()
op_slice_update()
op_stop_gradient()
op_switch()
op_unstack()
op_vectorized_map()
op_while_loop()

Other ops:
op_abs()
op_add()
op_all()
op_any()
op_append()
op_arange()
op_arccos()
op_arccosh()
op_arcsin()
op_arcsinh()
op_arctan()
op_arctan2()
op_arctanh()
op_argmax()
op_argmin()
op_argpartition()
op_argsort()
op_array()
op_associative_scan()
op_average()
op_average_pool()
op_batch_normalization()
op_binary_crossentropy()
op_bincount()
op_broadcast_to()
op_cast()
op_categorical_crossentropy()
op_ceil()
op_cholesky()
op_clip()
op_concatenate()
op_cond()
op_conj()
op_conv()
op_conv_transpose()
op_convert_to_numpy()
op_convert_to_tensor()
op_copy()
op_correlate()
op_cos()
op_cosh()
op_count_nonzero()
op_cross()
op_ctc_decode()
op_ctc_loss()
op_cumprod()
op_cumsum()
op_custom_gradient()
op_depthwise_conv()
op_det()
op_diag()
op_diagonal()
op_diff()
op_digitize()
op_divide()
op_divide_no_nan()
op_dot()
op_dtype()
op_eig()
op_eigh()
op_einsum()
op_elu()
op_empty()
op_equal()
op_erf()
op_erfinv()
op_exp()
op_expand_dims()
op_expm1()
op_extract_sequences()
op_eye()
op_fft()
op_fft2()
op_flip()
op_floor()
op_floor_divide()
op_fori_loop()
op_full()
op_full_like()
op_gelu()
op_get_item()
op_greater()
op_greater_equal()
op_hard_sigmoid()
op_hard_silu()
op_hstack()
op_identity()
op_imag()
op_image_affine_transform()
op_image_crop()
op_image_extract_patches()
op_image_hsv_to_rgb()
op_image_map_coordinates()
op_image_pad()
op_image_resize()
op_image_rgb_to_grayscale()
op_image_rgb_to_hsv()
op_in_top_k()
op_inv()
op_irfft()
op_is_tensor()
op_isclose()
op_isfinite()
op_isinf()
op_isnan()
op_istft()
op_leaky_relu()
op_less()
op_less_equal()
op_linspace()
op_log()
op_log10()
op_log1p()
op_log2()
op_log_sigmoid()
op_log_softmax()
op_logaddexp()
op_logical_and()
op_logical_not()
op_logical_or()
op_logical_xor()
op_logspace()
op_logsumexp()
op_lstsq()
op_lu_factor()
op_map()
op_matmul()
op_max()
op_max_pool()
op_maximum()
op_mean()
op_median()
op_meshgrid()
op_min()
op_minimum()
op_mod()
op_moments()
op_moveaxis()
op_multi_hot()
op_multiply()
op_nan_to_num()
op_ndim()
op_negative()
op_nonzero()
op_norm()
op_normalize()
op_not_equal()
op_one_hot()
op_ones()
op_ones_like()
op_outer()
op_pad()
op_power()
op_prod()
op_psnr()
op_qr()
op_quantile()
op_ravel()
op_real()
op_reciprocal()
op_relu()
op_relu6()
op_repeat()
op_reshape()
op_rfft()
op_roll()
op_round()
op_rsqrt()
op_scan()
op_scatter()
op_searchsorted()
op_segment_max()
op_segment_sum()
op_select()
op_selu()
op_separable_conv()
op_shape()
op_sigmoid()
op_sign()
op_silu()
op_sin()
op_sinh()
op_size()
op_slice()
op_slice_update()
op_slogdet()
op_softmax()
op_softplus()
op_softsign()
op_solve()
op_solve_triangular()
op_sort()
op_sparse_categorical_crossentropy()
op_split()
op_sqrt()
op_square()
op_squeeze()
op_stack()
op_std()
op_stft()
op_stop_gradient()
op_subtract()
op_sum()
op_svd()
op_swapaxes()
op_switch()
op_take()
op_take_along_axis()
op_tan()
op_tanh()
op_tensordot()
op_tile()
op_top_k()
op_trace()
op_transpose()
op_tri()
op_tril()
op_triu()
op_unstack()
op_var()
op_vdot()
op_vectorize()
op_vectorized_map()
op_vstack()
op_where()
op_while_loop()
op_zeros()
op_zeros_like()