Automated predictive modeling and framework
Systems and methods for providing a predictive framework are provided. The predictive framework comprises plural neural layers of adaptable, executable neurons. Neurons accept one or more input signals and produce an output signal that may be used by an upper-level neural layer. Input signals are received by an encoding neural layer, where there is a 1:1 correspondence between an input signal and an encoding neuron. Input signals for a set of data are received at the encoding layer and processed successively by the plurality of neural layers. An objective function utilizes the output signals of the topmost neural layer to generate predictive results for the data set according to an objective. In one embodiment, the objective is to determine the likelihood of user interaction with regard to a specific item of content in a set of search results, or the likelihood of user interaction with regard to any item of content in a set of search results.
1. A non-transitory computer-readable medium comprising executable instructions that, when executed by at least one processor of a machine, cause the machine to implement a computer-implemented framework for providing predicted results, comprising:
a plurality of neural layers comprising:
an encoding layer comprising a plurality of encoding neurons each having an input for receiving a corresponding input signal from a set of input data to evaluate and an output comprising a signal embedding for the input signal;
a plurality of residual layers each comprising a plurality of residual neurons, each residual neuron comprising a plurality of rectified linear operations without convolutional operations and a summation operation to add back elements of an input to the residual neuron, one of the plurality of residual layers being connected to the encoding layer and the other residual layers being connected to at least one other residual layer; and
an objective function that, in execution, determines predicted results from output signals of a topmost residual layer according to a predetermined objective;
wherein, in execution, the framework obtains the input signals by way of the encoding layer and processes the input signals successively through the plurality of neural layers to the topmost residual layer; and
wherein the objective function determines the predicted results from the output signals of the topmost residual layer according to the predetermined objective and provides the predicted results to a requesting party.
2. The medium of claim 1 , wherein the input signals comprise one or more of:
user identification;
query;
subject matter of a query;
entity of a query;
keywords;
advertiser;
advertisement campaign;
time of day;
day of week;
holiday;
season;
gender of a user, and
advertisement.
3. The medium of claim 1 , wherein the predetermined objective of the objective function is to determine a probability of user interaction with regard to a specific item of content in a set of search results.
4. The medium of claim 3 , wherein the specific item of content in the set of search results comprises an advertisement in the set of search results.
5. The medium of claim 3 , wherein the specific item of content in the set of search results comprises a sponsored search result in the set of search results.
6. The medium of claim 1 , wherein the predetermined objective of the objective function is to determine a probability of user interaction with regard to any item of content in a set of search results.
7. A computer system configured to generate predicted results with regard to base data, the computer system comprising a processor and a memory, wherein the processor executes instructions stored in the memory as part of or in conjunction with additional executable components to generate the predicted results, comprising:
an executable predictive framework, wherein the predictive framework comprises:
a plurality of neural layers comprising an encoding layer and a plurality of residual layers, wherein:
each neural layer comprises a plurality of neurons, each neuron comprising an executable object that accepts one or more inputs from a lower layer and generates an output; and
the encoding layer is a lowest neural layer comprising a plurality of encoding neurons having a 1:1 correspondence with a plurality of input signals;
the plurality of residual layers comprising a plurality of residual neurons, each residual neuron comprising a plurality of rectified linear operations without convolutional operations and a summation operation to add back elements of an input to the residual neuron, one of the plurality of residual layers being connected to the encoding layer and the other residual layers being connected to at least one other residual layer; and
an objective function that, in execution, determines predicted results from output signals of a topmost residual layer according to a predetermined objective;
wherein, in execution, the framework obtains the input signals by way of the encoding layer and processes the input signals successively through the plurality of neural layers to the topmost residual layer; and
wherein the objective function determines the predicted results from the output signals of the topmost residual layer according to the predetermined objective and provides the predicted results to a requesting party.
8. The computer system of claim 7 , wherein the input signals comprise one or more of:
user identification;
query;
subject matter of a query;
entity of a query;
keywords;
advertiser;
advertisement campaign;
time of day;
day of week;
holiday;
season;
gender of a user, and
advertisement.
9. The computer system of claim 7 , wherein the predetermined objective of the objective function is to determine a likelihood of user interaction with regard to a specific item of content in a set of search results.
10. The computer system of claim 7 , wherein the predetermined objective of the objective function is to determine a likelihood of user interaction with regard to any item of content in a set of search results.
11. A computer-implemented method for determining predicted results for a set of data, the method comprising:
providing an executable predictive framework having a validated model, the executable predictive framework comprising:
a plurality of neural layers comprising a plurality of residual layers and an encoding layer, wherein:
each neural layer comprises a plurality of neurons, each neuron comprising an executable object that accepts one or more inputs and generates an output; and
the encoding layer is a lowest neural layer and comprises a plurality of encoding neurons having a 1:1 correspondence with a plurality of input signals;
each residual layer residing above the encoding layer, one of the plurality of residual layers being connected to the encoding layer and the other residual layers being connected to at least one other residual layer, each residual layer comprising a plurality of residual neurons, each residual neuron comprising a plurality of rectified linear operations without convolutional operations and a summation operation to add back elements of an input to the residual neuron; and
an objective function that, in execution, determines the predicted results from output signals of a topmost residual layer according to a predetermined objective;
wherein, in execution, the framework obtains the input signals by way of the encoding layer and processes the input signals successively through the plurality of neural layers to the topmost residual layer; and
wherein the objective function determines the predicted results from the output signals of the topmost residual layer according to the predetermined objective and provides the predicted results to a requesting party;
obtaining the input signals for processing by the predictive framework;
processing the input signals by the predictive framework:
obtaining the predicted results from the objective function; and
providing the predicted results to the requesting party.
12. The computer-implemented method of claim 11 , wherein the predetermined objective of the objective function is to determine a probability of user interaction with regard to a specific item of content in a set of search results.