HYPE: A Benchmark for Human eYe Perceptual Evaluation of Generative Models

Human evaluation for generative models have been ad-hoc.

They propose a standard human benchmark for generative realism that is grounded in psychophysics research in perception.

https://arxiv.org/abs/1904.01121
https://hype.stanford.edu/

BagNet: Berkeley Analog Generator with Layout Optimizer Boosted with Deep Neural Networks
Hakhamaneshi et al.: https://arxiv.org/abs/1907.10515
#SignalProcessing #MachineLearning #NeuralComputing

Equilibrium Propagation: Bridging the Gap between Energy-Based Models and Backpropagation


https://www.frontiersin.org/articles/10.3389/fncom.2017.00024/full

Intelligent artificial agents learning to play 'hide and seek'

https://www.profillic.com/paper/arxiv:1909.07528

rlpyt: A Research Code Base for Deep Reinforcement Learning in PyTorch
Blog by Adam Stookei : https://bair.berkeley.edu/blog/2019/09/24/rlpyt/
#DeepLearning #ReinforcementLearning #PyTorch

Deep Dynamics Models for Learning Dexterous Manipulation
Nagabandi et al.: https://arxiv.org/abs/1909.11652
#Robotics #MachineLearning #ReinforcementLearning

Identifying and eliminating bugs in learned predictive models
https://deepmind.com/blog/article/robust-and-verified-ai

Deep Learning model that good at Calculus. Nice.


https://openreview.net/forum?id=S1eZYeHFDS

AI solution about RealEstate in Singapore
They built on open-source geospatial features and were able to predict Singapore real estate prices with 87% accuracy (i.e., within an error margin of S$100).
They used a) OpenStreetMap (https://download.geofabrik.de/asia/malaysia-singapore-brunei.html)
b) Geomancer for geospatial features (https://stories.thinkingmachin.es/geomancer/)
Link: https://download.geofabrik.de/asia/malaysia-singapore-brunei.html

Extreme Language Model Compression with Optimal Subwords and Shared Projections
Zhao et al.: https://arxiv.org/abs/1909.11687
#neuralnetwork #bert #nlp

Mathematical Reasoning in Latent Space
Lee et al.: https://arxiv.org/pdf/1909.11851v1.pdf
#Mathematics #Reasoning #LatentSpace

Learning Pixel Representations for Generic Segmentation. https://arxiv.org/abs/1909.11735

Revisit Knowledge Distillation: a Teacher-free Framework. https://arxiv.org/abs/1909.11723

Augmenting the Pathology Lab: An Intelligent Whole Slide Image Classification System for... https://arxiv.org/abs/1909.11212

Scale-Equivariant Neural Networks with Decomposed Convolutional Filters. https://arxiv.org/abs/1909.11193

Neural reparameterization improves structural optimization
Hoyer et al.: https://arxiv.org/abs/1909.04240
#MachineLearning #NeuralNetworks #Optimization

Infinite-resolution image generation in the browser
Made by Erlend Gjesteland Ekern using CPPNs, GANs and TensorFlow.js : https://thispicturedoesnotexist.com
#CPPN #GAN #TensorflowJS

Representation Wars: Enacting an Armistice through Active Inference
https://www.researchgate.net/publication/333969028_Representation_Wars_Enacting_an_Armistice_through_Active_Inference

Artificial Design: Modeling Artificial Super Intelligence with Extended General Relativity and Universal Darwinism via Geometrization for Universal Design Automation
https://openreview.net/forum?id=SyxQ_TEFwS

GAMIN: An Adversarial Approach to Black-Box Model Inversion. https://arxiv.org/abs/1909.11835

#weekend_read
Check out this mind-blowing survey on HRL with some really-strong proven hypothesis. Maybe the best till date!
Paper-Title: WHY DOES HIERARCHY (SOMETIMES) WORK SO WELL IN REINFORCEMENT LEARNING?
Link to the paper: https://arxiv.org/pdf/1909.10618.pdf
#GoogleAI #UCB
Four Hypothesis: The four hypotheses may also be categorized as hierarchical training (H1 and H3) and hierarchical exploration (H2 and H4).
(H1) Temporally extended training. High-level actions correspond to multiple environment steps. To the high-level agent, episodes are effectively shorter. Thus, rewards are propagated faster and learning should improve.
(H2) Temporally extended exploration. Since high-level actions correspond to multiple environment steps, exploration in the high-level is mapped to environment exploration which is temporally correlated across steps. This way, an HRL agent explores the environment more efficiently. As a motivating example, the distribution associated with a random (Gaussian) walk is wider when the random noise is temporally correlated.
(H3) Semantic training. High-level actor and critic networks are trained with respect to semantically meaningful actions. These semantic actions are more correlated with future values, and thus easier to learn, compared to training with respect to the atomic actions of the environment. For example, in a robot navigation task it is easier to learn future values with respect to deltas in x-y coordinates rather than robot joint torques.
(H4) Semantic exploration. Exploration strategies (in the simplest case, random action noise) are applied to semantically meaningful actions and are thus more meaningful than the same strategies would be if applied to the atomic actions of the environment. For example, in a robot navigation task, it intuitively makes more sense to explore at the level of x-y coordinates rather than robot joint torques.
TL;DR: A large number of conclusions can be drawn based on empirical analysis. Here are few:-
In terms of the benefits of training, it is clear that training with respect to semantically meaningful abstract actions (H3) has a negligible effect on the success of HRL.
Moreover, temporally extended training (H1) is only important insofar as it enables the use of multi-step rewards, as opposed to training with respect to temporally extended actions.
The main and arguably most surprising, the benefit of the hierarchy is due to exploration. This is evidenced by the fact that temporally extended goal-reaching and agent-switching can enable non-hierarchical agents to solve tasks that otherwise can only be solved.
These results suggest that the empirical effectiveness of hierarchical agents simply reflects the improved exploration that these agents can attain.