Fully Connected

Bringing ML practitioners together

Fully Connected is your home for curated tutorials, conversations with industry leaders, deep dives into the newest ML research, and a whole lot more.

FeaturedBlog PostsEventsGradient DissentForumComputer VisionNLPReinforcement LearningPyTorchKeras or TensorFlowfastaiHugging FaceScikitKaggleTwo Minute PapersReproducibility ChallengePaper OverviewVisualizationTutorialQuestionDiscussionResourcesShowcaseWebinarAMAPaper Reading GroupSalonfastbookCustom Charts
iTAML: An Incremental Task-Agnostic Meta-learning Approach
Joel Joseph
Meta-Consolidation for Continual Learning (MERLIN)
Shambhavi Mishra
JSTASR: Joint Size and Transparency-Aware Snow Removal Algorithm
Kajal Puri
Uncertainty-Guided Continual Learning With Bayesian Neural Networks
Shravan Nayak
Backpropagated Gradient Representations for Anomaly Detection
Shambhavi Mishra
Dynamic Group Convolution for Accelerating Convolutional Neural Networks (ECCV 2020)
Sachin Malhotra
Dynamic Convolutions: Exploiting Spatial Sparsity for Faster Inference (CVPR 2020)
Sachin Malhotra
Rethinking Depthwise Separable Convolutions: How Intra-Kernel Correlations Lead to Improved MobileNets (CVPR 2020)
Abhay Puri
Latent Embedding Feedback and Discriminative Features for Zero-Shot Classification
Animesh Gupta
Bad Global Minima Exist and SGD Can Reach Them
Gustavo Sutter P. Carvalho, João Araújo, João Pedro Rodrigues Mattos, João Marcos Cardoso da Silva, Bruno Gomes Coelho
When Does Self-Supervision Improve Few-Shot Learning?
Arjun Ashok, Haswanth Aekula
Meta Dropout: Learning to Perturb Latent Features for Better Generalization
Joel Joseph
SF-Net: Single-Frame Supervision for Temporal Action Localization
Kajal Puri
Funnel Activation for Visual Recognition (CVPR 2020)
Manoj Manjhari
Gated Channel Transformation for Visual Recognition (CVPR 2020)
Manoj Manjhari
Dynamic R-CNN: Towards High Quality Object Detection via Dynamic Training (ECCV 2020)
Strip Pooling: Rethinking Spatial Pooling for Scene Parsing (CVPR 2020)
Manoj Manjhari
ML Reproducibility Challenge Spring 2021
Diganta Misra
Interventional Few-Shot Learning
Lukasz Bala
Temporal Spike Sequence Learning via Backpropagation for Deep Spiking Neural Networks
Laura R Edmondson, Anil Ozdemir
On the Relationship Between Self-Attention and Convolutional Layers
Nishant Prabhu, Mukund Varma T
ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks
Diganta Misra
La-MAML: Look-ahead Meta Learning for Continual Learning
Joel Joseph
Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space
Kevin Jablonka
Rigging the Lottery: Making All Tickets Winners
Rajat Vadiraj Dwaraknath, Varun Sundar
Reformer: The Efficient Transformer
Morgan, Hallvar Gisnås, Arto, tyoc213, Priyank, Sheik Mohamed Imran, Dean P
Reproducible Models with Weights & Biases
Diganta Misra