and as an in

Subfields: Natural Language Processing

Citations in the corpus

• Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks (2015, methodology): “We evaluate our method on the PASCAL VOC detection benchmarks [4], where RPNs with Fast R-CNNs produce detection accuracy better than the strong baseline of Selective Search with Fast R-CNNs.”
• The Capacity for Moral Self-Correction in Large Language Models (2023, background): “” [28]–which is also related to prompting [57, 60] or training [39] models to “show their work”.”
• Data-Efficient Image Recognition with Contrastive Predictive Coding (2019, background): “The idea of semi-supervised learning is to leverage unlabeled data to improve task performance when few labeled examples are available [10, 70].”
• CCNet: Criss-Cross Attention for Semantic Segmentation (2018, methodology): “Most of methods [6, 21, 41, 27, 31, 42, 36, 19, 43, 37] adopt the same backbone as ours and the others [33, 35] utilize stronger backbones.”
• Contrastive Multiview Coding (2019, methodology): “The prior works most similar to our own (and inspira-tional to us) are Contrastive Predictive Coding (CPC) [46] and Deep InfoMax [22].”
• Towards VQA Models That Can Read (2019, methodology): “[45] introduced a Fact-Based VQA dataset which explicitly requires external knowledge to answer a question.”
• Bag of Tricks for Image Classification with Convolutional Neural Networks (2018, neutral): “We evaluate performance using PASCAL VOC [4].”
• Context Encoders: Feature Learning by Inpainting (2016, methodology): “We experiment with images from two datasets: Paris StreetView [8] and ImageNet [37] without using any of the accompanying labels.”
• Unsupervised Visual Representation Learning by Context Prediction (2015, methodology): “To ensure that our algorithm has not simply learned an object-centric representation due to the various biases [52] in ImageNet, we also applied our algorithm to 15,000 Street View images from Paris (following [13]).”
• Shape-aware Instance Segmentation (2016, methodology): “Instance-level semantic segmentation, which aims at jointly detecting, segmenting and classifying every individual object in an image, has recently become a core challenge in scene understanding [4, 21, 8].”
• Deformable part models are convolutional neural networks (2014, methodology): “We conducted several preliminary experiments on the PASCAL VOC 2007 dataset [6] in order to find a DPM structure suited to conv5 features.”
• The Unreasonable Effectiveness of Deep Features as a Perceptual Metric (2018, methodology): “We compute area under the curve, known as mAP [14].”
• The Reversible Residual Network: Backpropagation Without Storing Activations (2017, background): “ResNets currently represent the state-of-the-art in object recognition [13], semantic segmentation [35] and image generation [32].”
• Split-Brain Autoencoders: Unsupervised Learning by Cross-Channel Prediction (2016, methodology): “Table 4: Task and Dataset Generalization on PASCAL VOC Classification and detection on PASCAL VOC 2007 [13] and segmentation on PASCAL VOC 2012 [14], using mean average precision (mAP) and mean intersection over union (mIU) metrics for each task, with publicly available testing frameworks from [25],”
• Controlled Sparsity via Constrained Optimization or: How I Learned to Stop Tuning Penalties and Love Constraints (2022, background): “fairness [20, 5]) by prescribing them as additional constraints; much like extra additive terms in the penalized formulation.”
• A literature survey of low‐rank tensor approximation techniques (2013, neutral): “The combination of ALS with the TT decomposition has been considered in [65–67 , The ALS scheme can be improved in various ways.”
• Global Pooling, More than Meets the Eye: Position Information is Encoded Channel-Wise in CNNs (2021, positive): “Decoding absolute location from ImageNet [4] pretrained classification and PASCAL VOC 2012 [5] pretrained segmentation models (ResNet-18 [7] backbone) using GAPNet and ShuffleNet.”
• Deep Extreme Cut: From Extreme Points to Object Segmentation (2017, methodology): “Instead, we train our network by taking advantage of already existing segmentation datasets [8, 21], by recovering the extreme points from the masks of already annotated objects.”
• Bottom-Up and Top-Down Attention for Image Captioning and Visual Question Answering (2017, background): “[3] for further details of the VQA model implementation.”