18384586. MULTI-TURN DIALOGUE RESPONSE GENERATION USING ASYMMETRIC ADVERSARIAL MACHINE CLASSIFIERS simplified abstract (Capital One Services, LLC)
MULTI-TURN DIALOGUE RESPONSE GENERATION USING ASYMMETRIC ADVERSARIAL MACHINE CLASSIFIERS
Organization Name
Inventor(s)
Oluwatobi Olabiyi of Arlington VA (US)
Erik T. Mueller of Chevy Chase MD (US)
MULTI-TURN DIALOGUE RESPONSE GENERATION USING ASYMMETRIC ADVERSARIAL MACHINE CLASSIFIERS - A simplified explanation of the abstract
This abstract first appeared for US patent application 18384586 titled 'MULTI-TURN DIALOGUE RESPONSE GENERATION USING ASYMMETRIC ADVERSARIAL MACHINE CLASSIFIERS
Simplified Explanation
In a variety of embodiments, machine classifiers may model multi-turn dialogue as a one-to-many prediction task. The machine classifier may be trained using adversarial bootstrapping between a generator and a discriminator with multi-turn capabilities. The machine classifiers may be trained in both auto-regressive and traditional teacher-forcing modes, with the generator including a hierarchical recurrent encoder-decoder network and the discriminator including a bi-directional recurrent neural network. The discriminator input may include a mixture of ground truth labels, the teacher-forcing outputs of the generator, and/or noise data. This mixture of input data may allow for richer feedback on the autoregressive outputs of the generator. The outputs can be ranked based on the discriminator feedback and a response selected from the ranked outputs.
- Machine classifiers model multi-turn dialogue as a one-to-many prediction task.
- Training involves adversarial bootstrapping between a generator and a discriminator.
- Auto-regressive and traditional teacher-forcing modes are used for training.
- Generator includes a hierarchical recurrent encoder-decoder network.
- Discriminator includes a bi-directional recurrent neural network.
- Discriminator input includes ground truth labels, generator outputs, and noise data.
- Outputs are ranked based on discriminator feedback for response selection.
---
- Potential Applications
- Chatbots for customer service
- Virtual assistants for natural language interaction
- Automated messaging systems for communication platforms
- Problems Solved
- Improving the quality of responses in multi-turn dialogue systems
- Enhancing the conversational abilities of AI systems
- Providing more engaging and interactive interactions with users
- Benefits
- More accurate and contextually relevant responses
- Better understanding and interpretation of user input
- Enhanced user experience in conversational AI applications
Original Abstract Submitted
In a variety of embodiments, machine classifiers may model multi-turn dialogue as a one-to-many prediction task. The machine classifier may be trained using adversarial bootstrapping between a generator and a discriminator with multi-turn capabilities. The machine classifiers may be trained in both auto-regressive and traditional teacher-forcing modes, with the generator including a hierarchical recurrent encoder-decoder network and the discriminator including a bi-directional recurrent neural network. The discriminator input may include a mixture of ground truth labels, the teacher-forcing outputs of the generator, and/or noise data. This mixture of input data may allow for richer feedback on the autoregressive outputs of the generator. The outputs can be ranked based on the discriminator feedback and a response selected from the ranked outputs.