SparseKernelExpander

On-the-fly polynomial kernel expansion of sparse input.

This will perform polynomial kernel expansion of a set of sparse features based on a group attribute of this features. The features fed to this layer need to be a SparseFG[]. It assumes that the feature ids are uniform random numbers (hashes) so that we can efficiently generate a cross-feature hash by just XOR-ing together the single hashes of the monomial. This layer is meant to be used as part of a NeuralNet() topology at test time, but it's preferable to run the expansion outside the net at training time so that it can be run only once while building the dataset. This will avoid rebuilding the cross-features at every pass during training.

Constructors

this this(): Undocumented in source.
this this(ulong dim_out, string cross_feats_str, uint max_group_id, uint buff_single_feats_sz): Instantiate a new SparseKernelExpander layer.

Members

Functions

_init void _init(): Undocumented in source. Be warned that the author may not have intended to support it.
deserialize void deserialize(Serializer s): Undocumented in source. Be warned that the author may not have intended to support it.
dup NeuralLayer dup(): Undocumented in source. Be warned that the author may not have intended to support it.
expand void expand(T buff): Undocumented in source. Be warned that the author may not have intended to support it.
monitor void monitor(ushort id, uint hash, float val): Undocumented in source. Be warned that the author may not have intended to support it.
predict void predict(): Undocumented in source. Be warned that the author may not have intended to support it.
reset void reset(): Undocumented in source. Be warned that the author may not have intended to support it.
serialize void serialize(Serializer s): Undocumented in source. Be warned that the author may not have intended to support it.
share_params void share_params(NeuralLayer l): Undocumented in source. Be warned that the author may not have intended to support it.

Mixins

__anonymous mixin opCallNew: Undocumented in source.

Variables

_buff_single_feats_sz uint _buff_single_feats_sz;: Undocumented in source.
_cross_features_str string _cross_features_str;: Undocumented in source.
_max_group_id uint _max_group_id;: Undocumented in source.
cross2build ushort[][] cross2build;: Undocumented in source.
hash_buff uint[] hash_buff;: Undocumented in source.
here short[] here;: Undocumented in source.
single_hashes uint[][] single_hashes;: Undocumented in source.
single_vals float[][] single_vals;: Undocumented in source.
val_buff float[] val_buff;: Undocumented in source.

Mixed In Members

From `mixin opCallNew`

opCall auto opCall(T args): Undocumented in source. Be warned that the author may not have intended to support it.

Inherited Members

From InputLayer

input Variant input;: Undocumented in source.
forward_prop void forward_prop(T obs): Undocumented in source. Be warned that the author may not have intended to support it.
predict void predict(): Undocumented in source.
accumulate_grad void accumulate_grad(V[] grad): Undocumented in source. Be warned that the author may not have intended to support it.
backward_prop void backward_prop(V[] grad): Undocumented in source. Be warned that the author may not have intended to support it.
num_params ulong num_params [@property getter]: Undocumented in source. Be warned that the author may not have intended to support it.
recompute_topology void recompute_topology(): Undocumented in source. Be warned that the author may not have intended to support it.
allocate_interface void allocate_interface(): Undocumented in source. Be warned that the author may not have intended to support it.

Examples

// polynomial kernel (x_1 * x_3, x_1 * x_2 * x_4) in a 1k dimensional space:
auto l = SparseKernelExpander(1_000, "1^3,1^2^4");

SparseKernelExpander

Constructors

Members

Functions

Mixins

Variables

Mixed In Members

From `mixin opCallNew`

Inherited Members

From InputLayer

Examples

Meta

Source

SparseKernelExpander

Constructors

Members

Functions

Mixins

Variables

Mixed In Members

From mixin opCallNew

Inherited Members

From InputLayer

Examples

Meta

Source

From `mixin opCallNew`