Improving JEPAs with Spatial Conditioning: Robust and Efficient Representation Learning
This article was accepted into the Self-Supervised Learning Workshop: Theory and Practice (SSLTP) at NeurIPS 2024. The Image-based Joint Embedding ...
Tag: Representation
Google DeepMind researchers propose RT-Affordance: a hierarchical method that uses benefits as an intermediate representation of policies
In recent years, there has been significant development in the field of large pre-trained models for robot policy learning. The ...
REPLACEMENT: Untrained Representation Engineering for Managing Knowledge Conflicts in Large Language Models
Large language models (LLMs) have demonstrated impressive capabilities in handling knowledge-intensive tasks through their parametric knowledge stored within the model ...
Differentiable Robot Representation (Dr. Robot): A differentiable robot self-model from its visual appearance to its control parameters
Visual and action data are interconnected in robotic tasks, forming a perception-action loop. Robots rely on control parameters for movement, ...
Apple researchers propose a new AI algorithm to optimize a byte-level representation for automatic speech recognition (ASR) and compare it to the UTF-8 representation
End-to-end (E2E) neural networks have emerged as flexible and accurate models for multilingual automatic speech recognition (ASR). However, as the ...
Optimizing byte-level representation for end-to-end ASR
This paper was accepted into the IEEE Spoken Language technology (SLT) 2024 Workshop. In this paper, we propose an algorithm ...
Microsoft Releases Florence-2: A Novel Vision Foundation Model with a Unified Cue-Based Representation for a Variety of Computer Vision and Vision Language Tasks
There has been a marked movement in the field of AGI systems toward the use of adaptive, pretrained representations known ...
Embed Pose Graph, enabling 3D base model capabilities with a compact representation
This paper presents Embedding Pose Graph (EPG), an innovative method that combines the strengths of basic models with a simple ...
MOFI: Learning Image Representation from Annotated Images of Noisy Entities
In this paper, we present a novel approach to automatically assign entity labels to images from existing noisy image-text pairs. ...
The Representation Capability of LM Transformative Language Models with LM N-gram Language Models: Capturing the Parallelizable Nature of N-gram LMs
Neural language models (LMs) have become popular due to their extensive theoretical work mainly focused on representation ability. A previous ...
Recommended.
