Introduces causal inference-based bias correction for recommendation evaluation and learning, using propensity-weighted estimators and scalable matrix factorization.

Tobias Schnabel, Adith Swaminathan, Ashudeep Singh et al.

Proposes Equilibrium Propagation, a biologically plausible energy-based learning algorithm that estimates gradients via local perturbations, enabling training of recurrent neural networks without explicit backpropagation.

Benjamin Scellier, Yoshua Bengio

LIME algorithm explains any classifier's predictions using locally interpretable models, enhancing model trustworthiness.

Marco Tulio Ribeiro, Sameer Singh, Carlos Guestrin

Proposes asynchronous deep reinforcement learning (A3C) framework, outperforming DQN on Atari with half the training time on CPU.

Volodymyr Mnih, Adrià Puigdomènech Badia, Mehdi Mirza et al.

Recurrent neural network-based session recommender, outperforming traditional methods with 20%+ improvements in Recall@20.

Balázs Hidasi, Alexandros Karatzoglou, Linas Baltrunas et al.

Neural Programmer-Interpreter (NPI) combines LSTM core, persistent program memory, and environment encoders to enable multi-task program learning and generalization.

Scott Reed, Nando de Freitas

Geometric analysis reveals that wider neural networks have higher probability of initializing in basins with low objective values, facilitating optimization.

Itay Safran, Ohad Shamir

Diffusion-Convolutional Neural Networks (DCNN) leverage graph diffusion matrices for node classification, achieving 86-90% accuracy on benchmark datasets.

James Atwood, Don Towsley

BinaryConnect trains DNNs with binary weights during propagation, reducing multiplications to simple additions, achieving near state-of-the-art accuracy on MNIST, CIFAR-10, SVHN.

Matthieu Courbariaux, Yoshua Bengio, Jean-Pierre David

Deep spatial autoencoders extract environment feature points for visuomotor learning, combined with trajectory optimization for robotic control.

Chelsea Finn, Xin Yu Tan, Yan Duan et al.

Deep deterministic policy gradient (DDPG) enables stable, end-to-end reinforcement learning in high-dimensional continuous action spaces using deep neural networks.

Timothy P. Lillicrap, Jonathan J. Hunt, Alexander Pritzel et al.

Proposes a Wasserstein distance-based loss for multi-label learning, utilizing entropic regularization for efficient approximation, enhancing semantic smoothness.

Charlie Frogner, Chiyuan Zhang, Hossein Mobahi et al.

Proposes spectral networks with graph estimation for non-Euclidean data, matching or surpassing Dropout networks with fewer parameters.

Mikael Henaff, Joan Bruna, Yann LeCun

Permutation search is effective, but PP-Index/MI-file are not always faster than direct search.

Bilegsaikhan Naidan, Leonid Boytsov, Eric Nyberg

Proposes end-to-end training of deep visuomotor policies using Guided Policy Search with a 92,000-parameter CNN for direct image-to-torque mapping.

Sergey Levine, Chelsea Finn, Trevor Darrell et al.

K-FAC approximates Fisher matrix via Kronecker decomposition, enabling faster natural gradient optimization in neural networks.

James Martens, Roger Grosse

LINE efficiently embeds large-scale networks by optimizing first- and second-order proximities with edge sampling, handling millions of nodes and billions of edges.

Jian Tang, Meng Qu, Mingzhe Wang et al.

Exp3.G classifies feedback graphs: strongly observable gives ~√(αT), weakly observable ~δ^(1/3)T^(2/3), and unobservable Θ(T).

Noga Alon, Nicolò Cesa-Bianchi, Ofer Dekel et al.

Proposes a meta-algorithm (SAOL) transforming low-regret algorithms into strongly adaptive ones, ensuring near-optimal performance on every interval with \( O(\log T) \) overhead.

Amit Daniely, Alon Gonen, Shai Shalev-Shwartz

TRPO (Trust Region Policy Optimization) guarantees monotonic policy improvement using KL constraints, excelling in large neural network policy training for robotics and Atari games.

John Schulman, Sergey Levine, Philipp Moritz et al.