"NASiam: Efficient Representation Learning using Neural Architecture Search for Siamese Networks. (arXiv:2302.00059v1 [cs.CV])" — A novel approach that uses differentiable NAS to improve the multilayer perceptron projector and predictor (encoder/predictor pair) architectures inside siamese-networks-based contrastive learning frameworks (e.g., SimCLR, SimSiam, and MoCo) while preserving the simplicity of previous baselines.

Paper: http://arxiv.org/abs/2302.00059

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Layout of the NASiam architectu…

"Video Influencers: Unboxing the Mystique. (arXiv:2012.12311v2 [cs.LG] UPDATED)" — A study and analysis of YouTube influencers and their unstructured video data across text, audio and images using a novel "interpretable deep learning" framework to determine the effectiveness of their constituent elements in explaining video engagement.

Paper: http://arxiv.org/abs/2012.12311

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Gradient heat map in video fram…

"Attend-and-Excite: Attention-Based Semantic Guidance for Text-to-Image Diffusion Models. (arXiv:2301.13826v1 [cs.CV])" — A process which intervenes in the generative process of diffusion models on the fly during inference time to improve the faithfulness of the generated images to guide the model to refine the cross-attention units to attend to all subject tokens in the text prompt and strengthen - or excite - their activations, encouraging the model to generate all subjects described in the text prompt.

Paper: http://arxiv.org/abs/2301.13826
Code: https://github.com/AttendAndExcite/Attend-and-Excite

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Given a pre-trained text-to-ima…

"Grounding Language Models to Images for Multimodal Generation. (arXiv:2301.13823v1 [cs.CL])" — An efficient method to ground pretrained text-only language models to the visual domain, enabling them to process and generate arbitrarily interleaved image-and-text data. This approach apparently works with any off-the-shelf language model.

Paper: http://arxiv.org/abs/2301.13823

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Our method grounds a language m…

"UPop: Unified and Progressive Pruning for Compressing Vision-Language Transformers. (arXiv:2301.13741v1 [cs.CV])" — A universal vison-language Transformer compression framework which can handle multiple generative and discriminative vision-language tasks such as Visual Reasoning, Image Caption, Visual Question Answer, Image-Text Retrieval, Text-Image Retrieval, and Image Classification.

Paper: http://arxiv.org/abs/2301.13741

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Comparison between the Mask-bas…

"DisDiff: Unsupervised Disentanglement of Diffusion Probabilistic Models. (arXiv:2301.13721v1 [cs.CV])" — A new task to take advantage of the remarkable modeling ability of diffusion probabilistic models (DPM) using an unsupervised approach.

Paper: http://arxiv.org/abs/2301.13721

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Illustration of disentanglement…

"Learning Data Representations with Joint Diffusion Models. (arXiv:2301.13622v1 [cs.LG])" — A joint diffusion model that simultaneously learns meaningful internal representations fit for both generative and predictive tasks and which has superior performance across various tasks, including generative modeling, semi-supervised classification, and domain adaptation.

Paper: http://arxiv.org/abs/2301.13622

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Data representation zt in a UNe…

"NP-Match: Towards a New Probabilistic Model for Semi-Supervised Learning. (arXiv:2301.13569v1 [cs.CV])" — Adapting neural processes (NPs) for semi-supervised image classification tasks to arrive at a solution with much less computational overhead, which can save time at both the training and the testing phases.

Paper: http://arxiv.org/abs/2301.13569
Code: https://github.com/jianf-wang/np-match

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Overview of NP-Match: it contai…

"Domain-Generalizable Multiple-Domain Clustering. (arXiv:2301.13530v1 [cs.LG])" — Given unlabeled samples from multiple source domains, an attempt to learn a shared classifier that assigns the examples to various clusters by using the classifier for predicting cluster assignments in a previously unseen domain.

Paper: http://arxiv.org/abs/2301.13530

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Problem statement - given unlab…

"Fourier Sensitivity and Regularization of Computer Vision Models. (arXiv:2301.13514v1 [cs.CV])" — A study of the frequency sensitivity characteristics of deep neural networks using a principled approach due to recent work showing that deep neural networks latch on to the Fourier statistics of training data and show increased sensitivity to Fourier-basis directions in the input.

Paper: http://arxiv.org/abs/2301.13514

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Computing Fourier-sensitivity. …

"Conversational Automated Program Repair" — A method to help developers automatically generate patches for bugs using Large Language Models (LLMs) using a conversational approach for patch generation and validation.

Paper: https://arxiv.org/abs/2301.13246

#AI #NewPaper #MachineLearning #SoftwareEngineering

<<Find this useful? Please boost so that others can benefit too 🙂>>
Overview of conversational APR …

"Continuous Spatiotemporal Transformers. (arXiv:2301.13338v1 [cs.LG])" — A new transformer architecture that is designed for the modeling of continuous systems which guarantees a continuous and smooth output via optimization in Sobolev space.

Paper: http://arxiv.org/abs/2301.13338

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Diagram of CST’s workflow. (A) …

"ERA-Solver: Error-Robust Adams Solver for Fast Sampling of Diffusion Probabilistic Models. (arXiv:2301.12935v2 [cs.LG] UPDATED)" — An error-robust Adams solver (ERA-Solver), which utilizes the implicit Adams numerical method that consists of a predictor and a corrector.

Paper: http://arxiv.org/abs/2301.12935
Note: The PDF for this version of the paper is currently not available

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
We adopt the pretrained diffusi…

"PromptMix: Text-to-image diffusion models enhance the performance of lightweight networks. (arXiv:2301.12914v2 [cs.CV] UPDATED)" — A method for artificially boosting the size of existing datasets, that can be used to improve the performance of lightweight networks.

Paper: http://arxiv.org/abs/2301.12914

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Overview of PromptMix

"MOAT: Alternating Mobile Convolution and Attention Brings Strong Vision Models. (arXiv:2210.01820v2 [cs.CV] UPDATED)" — A family of neural networks that build on top of mobile convolution (i.e., inverted residual blocks) and attention which not only enhances the network representation capacity, but also produces better downsampled features.

Paper: http://arxiv.org/abs/2210.01820
Code: https://github.com/google-research/deeplab2

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Block comparison. (a) The MBCon…

"DAG: Depth-Aware Guidance with Denoising Diffusion Probabilistic Models. (arXiv:2212.08861v2 [cs.CV] UPDATED)" — A guidance method for diffusion models that uses estimated depth information derived from the rich intermediate representations of diffusion models.

Paper: http://arxiv.org/abs/2212.08861

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Qualitative comparisons of synt…

"Training-Free Structured Diffusion Guidance for Compositional Text-to-Image Synthesis. (arXiv:2212.05032v2 [cs.CV] UPDATED)" — Improving the compositional skills of text-to-image models; specifically, obtainining more accurate attribute binding and better image compositions by incorporating linguistic structures with the diffusion guidance process based on the controllable properties of manipulating cross-attention layers in diffusion-based models.

Paper: http://arxiv.org/abs/2212.05032
Code: https://github.com/weixi-feng/structured-diffusion-guidance

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Three challenging phenomena in …

"Refining Generative Process with Discriminator Guidance in Score-based Diffusion Models. (arXiv:2211.17091v2 [cs.CV] UPDATED)" — A generative SDE with score adjustment using an auxiliary discriminator with the goal of improving the original generative process of a pre-trained diffusion model by estimating the gap between the pre-trained score estimation and the true data score.

Paper: http://arxiv.org/abs/2211.17091

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Comparison of the denoising pro…

"Learning on tree architectures outperforms a convolutional feedforward network. (arXiv:2211.11378v3 [cs.CV] UPDATED)" — A 3-layer tree architecture inspired by experimental-based dendritic tree adaptations is developed and applied to the offline and online learning of the CIFAR-10 database to show that this architecture outperforms the achievable success rates of the 5-layer convolutional LeNet.

Paper: http://arxiv.org/abs/2211.11378

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Comparison of offline and onlin…

"Continual Learning by Modeling Intra-Class Variation. (arXiv:2210.05398v2 [cs.LG] UPDATED)" — An examination of memory-based continual learning which identifies that large variation in the representation space is crucial for avoiding catastrophic forgetting.

Paper: http://arxiv.org/abs/2210.05398
Code: https://github.com/yulonghui/moca

#AI #CV #NewPaper #DeepLearning #MachineLearning

<<Find this useful? Please boost so that others can benefit too 🙂>>
Average intra-class angle devia…