Visual Representations: Defining Properties and Deep Approximations

SVD-based Visualisation and Approximation for Time Series Data in Smart Energy Systems

Many time series in smart energy systems exhibit two different timescales. On the one hand there are patterns linked to daily human activities. On the other hand, there are relatively slow trends linked to seasonal variations. In this paper we interpret these time series as matrices, to be visualized as images. This approach has two advantages: First of all, interpreting such time series as images enables one to visually integrate across the image and makes it therefore easier to spot subtle or faint features. Second, the matrix interpretation also grants elucidation of the underlying structur…

CAPA：面向高效大视觉-语言模型的贡献感知剪枝与FFN近似

CAPA: Contribution-Aware Pruning and FFN Approximation for Efficient Large Vision-Language Models

CAPA通过贡献感知剪枝和FFN近似，提升大视觉-语言模型的效率与鲁棒性。

Approximated and User Steerable tSNE for Progressive Visual Analytics

Progressive Visual Analytics aims at improving the interactivity in existing analytics techniques by means of visualization as well as interaction with intermediate results. One key method for data analysis is dimensionality reduction, for example, to produce 2D embeddings that can be visualized and analyzed efficiently. t-Distributed Stochastic Neighbor Embedding (tSNE) is a well-suited technique for the visualization of several high-dimensional data. tSNE can create meaningful intermediate results but suffers from a slow initialization that constrains its application in Progressive Visual An…

Cherry-Picking Gradients: Learning Low-Rank Embeddings of Visual Data via Differentiable Cross-Approximation

We propose an end-to-end trainable framework that processes large-scale visual data tensors by looking at a fraction of their entries only. Our method combines a neural network encoder with a tensor train decomposition to learn a low-rank latent encoding, coupled with cross-approximation (CA) to learn the representation through a subset of the original samples. CA is an adaptive sampling algorithm that is native to tensor decompositions and avoids working with the full high-resolution data explicitly. Instead, it actively selects local representative samples that we fetch out-of-core and on-de…

基础模型定义视觉新纪元：综述与展望

Foundational Models Defining a New Era in Vision: A Survey and Outlook

本综述系统梳理了视觉基础模型，涵盖多模态架构、训练目标、预训练数据、微调及提示模式，并探讨了评估、理解、偏见等挑战，展望了未来研究方向。

Curriculum Learning for Compositional Visual Reasoning

Visual Question Answering (VQA) is a complex task requiring large datasets and expensive training. Neural Module Networks (NMN) first translate the question to a reasoning path, then follow that path to analyze the image and provide an answer. We propose an NMN method that relies on predefined cross-modal embeddings to ``warm start'' learning on the GQA dataset, then focus on Curriculum Learning (CL) as a way to improve training and make a better use of the data. Several difficulty criteria are employed for defining CL methods. We show that by an appropriate selection of the CL method the cost…

视觉自回归Transformer的通用逼近性

Universal Approximation of Visual Autoregressive Transformers

该研究扩展对Transformer基础模型基本限制的分析至视觉自回归（VAR）Transformer，证明单头单自注意力层单插值层的VAR Transformer是图像到图像Lipschitz函数的通用逼近器，为高效VAR策略提供设计原则。

A Simple Framework for Contrastive Learning of Visual Representations

This paper presents SimCLR: a simple framework for contrastive learning of visual representations. We simplify recently proposed contrastive self-supervised learning algorithms without requiring specialized architectures or a memory bank. In order to understand what enables the contrastive prediction tasks to learn useful representations, we systematically study the major components of our framework. We show that (1) composition of data augmentations plays a critical role in defining effective predictive tasks, (2) introducing a learnable nonlinear transformation between the representation and…