In Advances in Neural Information Processing Systems

By David S. Touretzky, M. Mozer, Michael E. Hasselmo, RegressionChristopher, I. K, WilliamsNeural, GroupAston, UniversityBirmingham, UniversityBirmingham. 2011

Subfields: ML Systems & Hardware

How researchers see this paper

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

In ML Systems & Hardware

Citations in the corpus

High-resolution de novo structure prediction from primary sequence (2022, background): “The recent success of DeepMind’s AlphaFold2 (1) and RoseTTAFold (2) in protein structure prediction has mainly been based on advances in deep learning, especially transformer-based models (3), and the accumulation of large databases of protein sequences and structures that enable effective training ”
Differentiable Physics and Stable Modes for Tool-Use and Manipulation Planning (2018, background): “Fully (auto-) differentiable physical simulations are currently discussed as candidates for inverting physics [Todorov, 2011; Filipe de Avila Belbute-Peres and J. Zico Kolter, 2017], and could be embedded in end-to-end trainable systems [Tamar et al. , 2016].”
Options Discovery with Budgeted Reinforcement Learning (2016, background): “In the computer science domain, these works have led to the hierarchical reinforcement learning paradigm [Dayan and Hinton, 1993; Dietterich, 1998; Parr and Russell, 1998], motivated by the idea that it makes the discovery of complex skills easier.”
Pay Attention to Features, Transfer Learn Faster CNNs (2020, methodology): “Learning without Forgetting (Li & Hoiem, 2018) learns to adapt the model to new tasks, while trying to match the output response on the original task of the original model using knowledge distillation (KD) (Hinton et al., 2014).”
The Dependence of Effective Planning Horizon on Model Accuracy (2015, methodology): “The dependence on the policy complexity ⇧R, is the novelty of our bound, compared to related work bounding loss by model errors or Bellman residuals [Kearns and Singh, 2002; Strehl et al., 2009; Farahmand et al., 2010].”
Human Gaze Assisted Artificial Intelligence: A Review (2020, background): “AI agents can learn to develop their own attention mechanism which is the key component of many state-of-the-art models [Mnih et al. , 2014; Vaswani et al. , 2017].”
GeoMAN: Multi-level Attention Networks for Geo-sensory Time Series Prediction (2018, methodology): “• Seq2seq [Sutskever et al., 2014]: It uses a RNN to encode the input sequences into a feature representation and another RNN to make predictions iteratively.”
Do Subsampled Newton Methods Work for High-Dimensional Data? (2019, methodology): “If the loss function of (1) is non-quadratic, the Hessian matrix Ht changes with iteration, and we can only guarantee fast local convergence, as well as the prior works [Roosta Khorasani and Mahoney, 2016; Xu et al., 2016].”