Modeling language learning for social cognition research: The effect of prior knowledge base subtleties – The goal of this paper is to establish and quantify how semantic representation of human language is affected by the presence of a wide variety of semantic entities. The aim of this paper is to present the concept of a new conceptual language to describe human language as an unstructured semantic space: it encompasses human objects, words, concepts and sentences. We believe that semantic representations of human language will help in exploring the domain of language ontology.

Recently, Convolutional neural networks (CNNs) and LSTMs are widely used to model many tasks. However, CNNs are expensive to train and test. This work proposes a novel approach inspired by deep Reinforcement Learning with Deep Belief Networks (LFLN). Our solution is using a LFLN for performing objective functions in CNNs, and then using a CNN for labeling the task (e.g., driving a car). We propose a novel deep learning technique based on the use of a CNN for labeling the task (e.g., detecting the driver’s intentions to drive). We show that an LFLN can be trained with a CNN for labeling the task (such as a human driver) and then it is possible to scale up the CNN to a deep LFLN. Experiment 2 showed that this approach achieved competitive results.

Probabilistic Modeling of Time-Series for Spatio-Temporal Data with a Bayesian Network Adversary

A note on the Lasso-dependent Latent Variable Model

Modeling language learning for social cognition research: The effect of prior knowledge base subtleties

Convex-constrained Inference with Structured Priors with Applications in Statistical Machine Learning and Data Mining

Leveraging Side Experiments with Deep Reinforcement Learning for Driver Test Speed PredictionRecently, Convolutional neural networks (CNNs) and LSTMs are widely used to model many tasks. However, CNNs are expensive to train and test. This work proposes a novel approach inspired by deep Reinforcement Learning with Deep Belief Networks (LFLN). Our solution is using a LFLN for performing objective functions in CNNs, and then using a CNN for labeling the task (e.g., driving a car). We propose a novel deep learning technique based on the use of a CNN for labeling the task (e.g., detecting the driver’s intentions to drive). We show that an LFLN can be trained with a CNN for labeling the task (such as a human driver) and then it is possible to scale up the CNN to a deep LFLN. Experiment 2 showed that this approach achieved competitive results.