DistriBlock: Identifying Adversarial Audio Samples for ASR Systems
DistriBlock proposes a novel detection strategy for identifying adversarial audio samples in ASR systems by analyzing output distribution characteristics.
Negating Negatives: Achieving Alignment with Human-Annotated Negative Samples
Distributional Dispreference Optimization (D2O) achieves alignment using solely human-annotated negative samples, reducing harmfulness while maintaining helpfulness.
Bi-level Learnable Large Language Model Planning for Long-Term Recommendation
Incorporating planning capabilities into recommendation systems enhances long-term engagement.
Negating Negatives: Achieving Alignment with Human-Annotated Negative Samples for Large Language Models
Proposing Distributional Dispreference Optimization (D2O) to achieve alignment using solely human-annotated negative samples, reducing harmfulness while maintaining helpfulness.
Understanding the Limitations of Explainable AI in Machine Learning
The author argues that existing accounts of scientific explanation cannot be effectively applied to deep neural networks, suggesting a shift towards "understandable AI" to avoid confusion and promote pragmatic understanding.
© 2024 by Linnk AI