Advances in Neural Information Processing Systems (NIPS)

By Vittorio Ferrari, Andrew Zisserman. 2007

707 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Evaluation & Benchmarks

Citations in the corpus

Order Matters: Sequence to sequence for sets (2015, neutral): “• A write block, which is an LSTM pointer network (Vinyals et al., 2015a) that takes in q ∗ T (as the context it needs from which to produce the output from the input set), and points at elements of m i (implicitly, x i ), one step at a time.”
Compositionality Decomposed: How do Neural Networks Generalise? (2019, background): “, 2017] and a transformer model (Transformer) [Vaswani et al., 2017].”
A Convergence Theory for Deep Learning via Over-Parameterization (2018, background): “We refer interested readers to some practical evidence [51, 54] that deeper (and wider) neural networks generalize better.”
Neural Programmer-Interpreters (2015, methodology): “Vinyals et al. (2015) developed Pointer Networks that generalize the notion of encoder attention in order to provide the decoder a variable-sized output space depending on the input sequence length.”
Deep Modular Co-Attention Networks for Visual Question Answering (2019, positive): “These features are the intermediate features extracted from a Faster R-CNN model (with ResNet-101 as its backbone) [26] pre-trained on the Visual Genome dataset [18].”
How Much Reading Does Reading Comprehension Require? A Critical Investigation of Popular Benchmarks (2018, methodology): “KV-MemNets are based on Memory Networks (Sukhbaatar et al., 2015), shown to perform well on both datasets.”
Data Poisoning against Differentially-Private Learners: Attacks and Defenses (2019, positive): “We are not the first to design defenses to data poisoning [Steinhardt et al., 2017; Raghunathan et al., 2018], but our proposal has several notable strengths compared to previous work.”
Reinforcement learning in robotics: A survey (2013, background): “Modeling exploration models with probability distributions has surprising implications, e.g. stochastic policies have been shown to be the optimal stationary policies for selected problems (Jaakkola et al., 1993; Sutton et al., 1999) and can even break the curse of dimensionality (Rust, 1997).”
Counterfactual Prediction with Deep Instrumental Variables Networks (2016, background): “(14) Indeed, this ‘two-draw’ ∇̂θLt is the optimal implementation of (13) in a large-scale learning setting [Bousquet and Bottou, 2008] where computation, rather than the amount of data available, is the binding constraint on error minimization.”
On the Equivalence between Kernel Quadrature Rules and Random Feature Expansions (2015, methodology): “For these problems, techniques based on positive deﬁnite kernels have emerged as having the potential to efﬁ-ciently deal with these situations, and to improve over plain Monte-Carlo integration (O’Hagan, 1991; Rasmussen and Ghahramani, 2003; Husz´ar and Duvenaud, 2012; Oates and Girolami, 2015).”
Recovery Guarantees for One-hidden-layer Neural Networks (2017, methodology): “Our idea follows the techniques used in (Zhong et al., 2016), where they ﬁrst used a 2nd-order moment P 2 to approximate the sub-space spanned by { w ∗ 1 , w ∗ 2 , · · · , w ∗ k } , denoted as V , then use V to reduce a higher-dimensional third-order tensor P 3 ∈ R d × d × d to a lower-dimensional…”
Monotonic Calibrated Interpolated Look-Up Tables (2015, background): “…kernel estimate post-process yes 2447 Ben-David (1995) tree penalize splits yes 125 Sill and Abu-Mostafa (1997) neural net penalize pairs no 550 Sill (1998) neural net constrain function yes 196 Kay and Ungar (2000) neural net constrain function yes 100 Potharst and Feelders (2002a) tree…”
Neural Generative Question Answering (2015, methodology): “TransE [Bordes et al. , 2013] is a method that learns the embedding vectors of the entities and the relations be-tween entities by translating from subject entities to object entities.”
Towards Vision-Based Deep Reinforcement Learning for Robotic Motion Control (2015, methodology): “The DQN used a deep convolutional neural network (CNN) [Krizhevsky et al., 2012] to approximate a Q-value function.”
DeViSE: A Deep Visual-Semantic Embedding Model (2013, methodology): “The DeViSE model was trained by asynchronous stochastic gradient descent on a distributed computing platform described in [4].”
Survey on distance metric learning and dimensionality reduction in data mining (2014, background): “Intuitively, active distance metric learning wants to select the most confusing data pairs such that the distance metric can be more effectively and efficiently learned. The authors in Yang et al. (2007) proposed a Bayesian approach called BADML to achieve such a goal.”