Hidden Markov model (HMM)
A Hidden Markov Model (HMM) is a statistical model that represents systems with hidden states. It is widely used in various applications such as speech recognition and bioinformatics.
Artificial Intelligence
Neural variational inference
Neural variational inference is a machine learning technique that uses neural networks to approximate complex posterior distributions in probabilistic models. It combines variational inference with deep learning to enable scalable and flexible inference in models where exact Bayesian inference is intractable.
Retrieval-augmented generation (RAG)
Retrieval-augmented generation (RAG) is a hybrid approach in natural language processing that combines information retrieval with generative models to produce more accurate and informative text. By integrating external knowledge sources, RAG enhances the capabilities of language models beyond their training data.
CLIFF (camera-LiDAR fusion for human mesh)
CLIFF (camera-LiDAR fusion for human mesh) is a computational approach that integrates camera imagery and LiDAR sensor data to reconstruct detailed 3D human body meshes. It utilizes complementary strengths of visual and depth information to enhance accuracy in human shape and pose estimation.
Universal adversarial perturbation
Universal adversarial perturbation refers to a single, small noise pattern that can be added to multiple inputs to fool machine learning models, especially deep neural networks, causing them to misclassify data across diverse samples.
Word error rate (WER)
Word error rate (WER) is a metric used to evaluate the performance of speech recognition systems by quantifying the differences between a recognized word sequence and a reference transcript. It is calculated based on the number of insertions, deletions, and substitutions needed to transform the hypothesis into the reference.
Capsule neural network
A capsule neural network is an advanced type of artificial neural network designed to improve the recognition of objects in images by preserving hierarchical relationships and spatial information. It addresses limitations of traditional convolutional neural networks by grouping neurons into capsules that capture pose and other properties.
Isomap
Isomap is a manifold learning technique used for dimensionality reduction while preserving geodesic distances between points in a dataset.
GLM (language model)
GLM (General Language Model) is a type of pre-trained language model developed primarily in China that supports both English and Chinese language understanding and generation tasks. It is designed to perform various natural language processing applications, including text generation, completion, and understanding.
SocialIQA
SocialIQA is a benchmark dataset designed to evaluate commonsense reasoning in artificial intelligence, focusing specifically on social situations and interpersonal interactions.