# Conference Poster Sessions

Below are the Conference Track papers presented at each of the poster sessions (on Monday, Tuesday or Wednesday, in the morning or evening). To find a paper, look for the poster with the corresponding number in the area dedicated to the Conference Track.

# Note to the Presenters

Each poster panel is 2 meters large and 1 meter tall.

If needed, tape will be provided to fix your poster.

### Monday Morning (April 24th, 10:30am to 12:30pm)

C1: Making Neural Programming Architectures Generalize via Recursion

C2: Learning Graphical State Transitions

C3: Distributed Second-Order Optimization using Kronecker-Factored Approximations

C4: Normalizing the Normalizers: Comparing and Extending Network Normalization Schemes

C5: Neural Program Lattices

C6: Diet Networks: Thin Parameters for Fat Genomics

C7: Unsupervised Cross-Domain Image Generation

C8: Towards Principled Methods for Training Generative Adversarial Networks

C9: Recurrent Mixture Density Network for Spatiotemporal Visual Attention

C10: Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfer

C11: Pruning Filters for Efficient ConvNets

C12: Stick-Breaking Variational Autoencoders

C13: Identity Matters in Deep Learning

C14: On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima

C15: Recurrent Hidden Semi-Markov Model

C16: Nonparametric Neural Networks

C17: Learning to Generate Samples from Noise through Infusion Training

C18: An Information-Theoretic Framework for Fast and Robust Unsupervised Learning via Neural Population Infomax

C19: Highway and Residual Networks learn Unrolled Iterative Estimation

C20: Soft Weight-Sharing for Neural Network Compression

C21: Snapshot Ensembles: Train 1, Get M for Free

C22: Towards a Neural Statistician

C23: Learning Curve Prediction with Bayesian Neural Networks

C24: Learning End-to-End Goal-Oriented Dialog

C25: Multi-Agent Cooperation and the Emergence of (Natural) Language

C26: Efficient Vector Representation for Documents through Corruption

C27: Improving Neural Language Models with a Continuous Cache

C28: Program Synthesis for Character Level Language Modeling

C29: Tracking the World State with Recurrent Entity Networks

C30: Reinforcement Learning with Unsupervised Auxiliary Tasks

C31: Neural Architecture Search with Reinforcement Learning

C32: Sample Efficient Actor-Critic with Experience Replay

C33: Learning to Act by Predicting the Future

### Monday Afternoon (April 24th, 4:30pm to 6:30pm)

C1: Neuro-Symbolic Program Synthesis

C2: Generative Models and Model Criticism via Optimized Maximum Mean Discrepancy

C3: Trained Ternary Quantization

C4: DSD: Dense-Sparse-Dense Training for Deep Neural Networks

C5: A Compositional Object-Based Approach to Learning Physical Dynamics

C6: Multilayer Recurrent Network Models of Primate Retinal Ganglion Cells

C7: Improving Generative Adversarial Networks with Denoising Feature Matching

C8: Transfer of View-manifold Learning to Similarity Perception of Novel Objects

C9: What does it take to generate natural textures?

C10: Emergence of foveal image sampling from learning to attend in visual scenes

C11: PixelCNN++: A PixelCNN Implementation with Discretized Logistic Mixture Likelihood and Other Modifications

C12: Learning to Optimize

C13: Do Deep Convolutional Nets Really Need to be Deep and Convolutional?

C14: Optimal Binary Autoencoding with Pairwise Correlations

C15: On the Quantitative Analysis of Decoder-Based Generative Models

C16: Adversarial machine learning at scale

C17: Transfer Learning for Sequence Tagging with Hierarchical Recurrent Networks

C18: Capacity and Learnability in Recurrent Neural Networks

C19: Deep Learning with Dynamic Computation Graphs

C20: Exploring Sparsity in Recurrent Neural Networks

C21: Structured Attention Networks

C22: Learning to Repeat: Fine Grained Action Repetition for Deep Reinforcement Learning

C23: Variational Lossy Autoencoder

C24: Learning to Query, Reason, and Answer Questions On Ambiguous Texts

C25: Deep Biaffine Attention for Neural Dependency Parsing

C26: A Compare-Aggregate Model for Matching Text Sequences

C27: Data Noising as Smoothing in Neural Network Language Models

C28: Neural Variational Inference For Topic Models

C29: Bidirectional Attention Flow for Machine Comprehension

C30: Q-Prop: Sample-Efficient Policy Gradient with An Off-Policy Critic

C31: Stochastic Neural Networks for Hierarchical Reinforcement Learning

C32: Learning Invariant Feature Spaces to Transfer Skills with Reinforcement Learning

C33: Third Person Imitation Learning

### Tuesday Morning (April 25th, 10:30am to 12:30pm)

C1: DeepDSL: A Compilation-based Domain-Specific Language for Deep Learning

C2: A SELF-ATTENTIVE SENTENCE EMBEDDING

C3: Deep Probabilistic Programming

C4: Lie-Access Neural Turing Machines

C5: Learning Features of Music From Scratch

C6: Mode Regularized Generative Adversarial Networks

C7: End-to-end Optimized Image Compression

C8: Variational Recurrent Adversarial Deep Domain Adaptation

C9: Steerable CNNs

C10: Deep Predictive Coding Networks for Video Prediction and Unsupervised Learning

C11: PixelVAE: A Latent Variable Model for Natural Images

C12: A recurrent neural network without chaos

C13: Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer

C14: Tree-structured decoding with doubly-recurrent neural networks

C15: Introspection:Accelerating Neural Network Training By Learning Weight Evolution

C16: Hyperband: Bandit-Based Configuration Evaluation for Hyperparameter Optimization

C17: Quasi-Recurrent Neural Networks

C18: Attend, Adapt and Transfer: Attentive Deep Architecture for Adaptive Transfer from multiple sources in the same domain

C19: A Baseline for Detecting Misclassified and Out-of-Distribution Examples in Neural Networks

C20: Trusting SVM for Piecewise Linear CNNs

C21: Maximum Entropy Flow Networks

C22: The Concrete Distribution: A Continuous Relaxation of Discrete Random Variables

C23: Unrolled Generative Adversarial Networks

C24: A Simple but Tough-to-Beat Baseline for Sentence Embeddings

C25: Query-Reduction Networks for Question Answering

C26: Machine Comprehension Using Match-LSTM and Answer Pointer

C27: Words or Characters? Fine-grained Gating for Reading Comprehension

C28: Dynamic Coattention Networks For Question Answering

C29: Multi-view Recurrent Neural Acoustic Word Embeddings

C30: Episodic Exploration for Deep Deterministic Policies for StarCraft Micromanagement

C31: Training Agent for First-Person Shooter Game with Actor-Critic Curriculum Learning

C32: Generalizing Skills with Semi-Supervised Reinforcement Learning

C33: Improving Policy Gradient by Exploring Under-appreciated Rewards

### Tuesday Afternoon (April 25th, 2:00pm to 4:00pm)

C1: Sigma Delta Quantized Networks

C2: Paleo: A Performance Model for Deep Neural Networks

C3: DeepCoder: Learning to Write Programs

C4: Topology and Geometry of Deep Rectified Network Optimization Landscapes

C5: Incremental Network Quantization: Towards Lossless CNNs with Low-precision Weights

C6: Learning to Perform Physics Experiments via Deep Reinforcement Learning

C7: Decomposing Motion and Content for Natural Video Sequence Prediction

C8: Calibrating Energy-based Generative Adversarial Networks

C9: Pruning Convolutional Neural Networks for Resource Efficient Inference

C10: Incorporating long-range consistency in CNN-based texture generation

C11: Lossy Image Compression with Compressive Autoencoders

C12: LR-GAN: Layered Recursive Generative Adversarial Networks for Image Generation

C13: Semi-supervised Knowledge Transfer for Deep Learning from Private Training Data

C14: Deep Variational Bayes Filters: Unsupervised Learning of State Space Models from Raw Data

C15: Mollifying Networks

C16: beta-VAE: Learning Basic Visual Concepts with a Constrained Variational Framework

C17: Categorical Reparameterization with Gumbel-Softmax

C18: Online Bayesian Transfer Learning for Sequential Data Modeling

C19: Latent Sequence Decompositions

C20: Density estimation using Real NVP

C21: Recurrent Batch Normalization

C22: SGDR: Stochastic Gradient Descent with Restarts

C23: Variable Computation in Recurrent Neural Networks

C24: Deep Variational Information Bottleneck

C25: SampleRNN: An Unconditional End-to-End Neural Audio Generation Model

C26: TopicRNN: A Recurrent Neural Network with Long-Range Semantic Dependency

C27: Frustratingly Short Attention Spans in Neural Language Modeling

C28: Offline Bilingual Word Vectors, Orthogonal Transformations and the Inverted Softmax

C29: LEARNING A NATURAL LANGUAGE INTERFACE WITH NEURAL PROGRAMMER

C30: Designing Neural Network Architectures using Reinforcement Learning

C31: Metacontrol for Adaptive Imagination-Based Optimization

C32: Recurrent Environment Simulators

C33: EPOpt: Learning Robust Neural Network Policies Using Model Ensembles

### Wednesday Morning (April 26th, 10:30am to 12:30pm)

C1: Deep Multi-task Representation Learning: A Tensor Factorisation Approach

C2: Training deep neural-networks using a noise adaptation layer

C3: Delving into Transferable Adversarial Examples and Black-box Attacks

C4: Towards the Limit of Network Quantization

C5: Towards Deep Interpretability (MUS-ROVER II): Learning Hierarchical Representations of Tonal Music

C6: Learning to superoptimize programs

C7: Regularizing CNNs with Locally Constrained Decorrelations

C8: Generative Multi-Adversarial Networks

C9: Visualizing Deep Neural Network Decisions: Prediction Difference Analysis

C10: FractalNet: Ultra-Deep Neural Networks without Residuals

C11: Faster CNNs with Direct Sparse Convolutions and Guided Pruning

C12: FILTER SHAPING FOR CONVOLUTIONAL NEURAL NETWORKS

C13: The Neural Noisy Channel

C14: Automatic Rule Extraction from Long Short Term Memory Networks

C15: Adversarially Learned Inference

C16: Deep Information Propagation

C17: Revisiting Classifier Two-Sample Tests

C18: Loss-aware Binarization of Deep Networks

C19: Energy-based Generative Adversarial Networks

C20: Central Moment Discrepancy (CMD) for Domain-Invariant Representation Learning

C21: Temporal Ensembling for Semi-Supervised Learning

C22: On Detecting Adversarial Perturbations

C23: Understanding deep learning requires rethinking generalization

C24: Adversarial Feature Learning

C25: Learning through Dialogue Interactions

C26: Learning to Compose Words into Sentences with Reinforcement Learning

C27: Batch Policy Gradient Methods for Improving Neural Conversation Models

C28: Tying Word Vectors and Word Classifiers: A Loss Framework for Language Modeling

C29: Geometry of Polysemy

C30: PGQ: Combining policy gradient and Q-learning

C31: Reinforcement Learning through Asynchronous Advantage Actor-Critic on a GPU

C32: Learning to Navigate in Complex Environments

C33: Learning and Policy Search in Stochastic Dynamical Systems with Bayesian Neural Networks

### Wednesday Afternoon (April 26th, 4:30pm to 6:30pm)

C1: Learning recurrent representations for hierarchical behavior modeling

C2: Predicting Medications from Diagnostic Codes with Recurrent Neural Networks

C3: Sparsely-Connected Neural Networks: Towards Efficient VLSI Implementation of Deep Neural Networks

C4: HolStep: A Machine Learning Dataset for Higher-order Logic Theorem Proving

C5: Learning Invariant Representations Of Planar Curves

C6: Entropy-SGD: Biasing Gradient Descent Into Wide Valleys

C7: Amortised MAP Inference for Image Super-resolution

C8: Inductive Bias of Deep Convolutional Networks through Pooling Geometry

C9: Neural Photo Editing with Introspective Adversarial Networks

C10: A Learned Representation For Artistic Style

C11: Learning to Remember Rare Events

C12: Optimization as a Model for Few-Shot Learning

C13: Support Regularized Sparse Coding and Its Fast Encoder

C14: Discrete Variational Autoencoders

C15: Training Compressed Fully-Connected Networks with a Density-Diversity Penalty

C16: Efficient Representation of Low-Dimensional Manifolds using Deep Networks

C17: Semi-Supervised Classification with Graph Convolutional Networks

C18: Understanding Neural Sparse Coding with Matrix Factorization

C19: Tighter bounds lead to improved classifiers

C20: Why Deep Neural Networks for Function Approximation?

C21: Hierarchical Multiscale Recurrent Neural Networks

C22: Dropout with Expectation-linear Regularization

C23: HyperNetworks

C24: Hadamard Product for Low-rank Bilinear Pooling

C25: Adversarial Training Methods for Semi-Supervised Text Classification

C26: Fine-grained Analysis of Sentence Embeddings Using Auxiliary Prediction Tasks

C27: Pointer Sentinel Mixture Models

C28: Reasoning with Memory Augmented Neural Networks for Language Comprehension

C29: Dialogue Learning With Human-in-the-Loop

C30: Zoneout: Regularizing RNNs by Randomly Preserving Hidden Activations

C31: Learning to Play in a Day: Faster Deep Reinforcement Learning by Optimality Tightening

C32: Learning Visual Servoing with Deep Features and Trust Region Fitted Q-Iteration

C33: An Actor-Critic Algorithm for Sequence Prediction