View source for 17958261. TRANSFER-LEARNING FOR STRUCTURED DATA WITH REGARD TO JOURNEYS DEFINED BY SETS OF ACTIONS simplified abstract (Microsoft Technology Licensing, LLC)

=TRANSFER-LEARNING FOR STRUCTURED DATA WITH REGARD TO JOURNEYS DEFINED BY SETS OF ACTIONS=


==Organization Name==

[[:Category:Microsoft Technology Licensing, LLC|Microsoft Technology Licensing, LLC]]


[[Category:Microsoft Technology Licensing, LLC]]

==Inventor(s)==

[[:Category:Kiran Rama of Bangalore (IN)|Kiran Rama of Bangalore (IN)]][[Category:Kiran Rama of Bangalore (IN)]]

[[:Category:Ke Li of Redmond WA (US)|Ke Li of Redmond WA (US)]][[Category:Ke Li of Redmond WA (US)]]

[[:Category:Sharath Kumar Rangappa of Bangalore (IN)|Sharath Kumar Rangappa of Bangalore (IN)]][[Category:Sharath Kumar Rangappa of Bangalore (IN)]]

[[:Category:Shariq Ahmad of Kolkata (IN)|Shariq Ahmad of Kolkata (IN)]][[Category:Shariq Ahmad of Kolkata (IN)]]

[[:Category:Akash Kodibail of Bengaluru (IN)|Akash Kodibail of Bengaluru (IN)]][[Category:Akash Kodibail of Bengaluru (IN)]]

==TRANSFER-LEARNING FOR STRUCTURED DATA WITH REGARD TO JOURNEYS DEFINED BY SETS OF ACTIONS - A simplified explanation of the abstract==


This abstract first appeared for US patent application 17958261 titled 'TRANSFER-LEARNING FOR STRUCTURED DATA WITH REGARD TO JOURNEYS DEFINED BY SETS OF ACTIONS

==Simplified Explanation==

The patent application describes techniques for transfer-learning for structured data related to journeys defined by sets of actions using deep neural networks.

* First DNN trained for a first journey using structured data.
* Weights of nodes in the first DNN transferred to nodes in a second DNN for a second journey.
* An embedding layer replaces the final layer of the first DNN in the second DNN.
* Weights of nodes in the embedding layer initialized based on the probability of co-occurrence of features.
* Softmax function applied on the final layer of the second DNN to indicate possible next actions.

== Potential Applications ==
The technology can be applied in various fields such as recommendation systems, personalized marketing, and predictive analytics.

== Problems Solved ==
This technology helps in improving the accuracy and efficiency of predicting next actions based on structured data related to journeys.

== Benefits ==
The benefits of this technology include enhanced decision-making, improved user experience, and increased automation in various processes.

== Potential Commercial Applications ==
Potential commercial applications include personalized product recommendations, targeted advertising, and optimized resource allocation in logistics.

== Possible Prior Art ==
One possible prior art could be the use of transfer learning in neural networks for pattern recognition tasks.

== Unanswered Questions ==

=== How does the embedding layer improve the transfer-learning process in the second DNN? ===
The embedding layer helps in capturing the relationships between features in the structured data, leading to better generalization and prediction performance.

=== What are the implications of using transfer-learning for structured data in real-time applications? ===
The real-time applications may require continuous updating of weights and embeddings based on new data, which could impact the overall performance and efficiency of the system.


==Original Abstract Submitted==

Techniques are described herein that are capable of performing transfer-learning for structured data with regard to journeys defined by sets of actions. A first deep neural network (DNN) for a first journey is trained using structured data. Weights of nodes in the first DNN are transferred to nodes in a second DNN for a second journey using transfer-learning. An embedding layer replaces a final layer of the first DNN in the second DNN to provide an output with a same number of nodes as a pre-final layer of the first DNN. Weights of the nodes in the embedding layer are initialized based at least on a probability that a new feature of the second journey co-occurs with each feature in the structured data. A softmax function is applied on a final layer of the second DNN to indicate possible next actions of the second journey.



[[Category:G06N3/08]]
[[Category:G06K9/62]]
[[Category:G06N3/04]]