gothamisland.xyz

Fast Learning using Conditional Explanations – This paper presents a new framework for solving a variety of optimization problems using conditional probability. We first show that a conditional probability loss can be learned in a general setting, and then use probabilistic inference to estimate the likelihood of the conditional probability loss, which are computationally efficient […]

The Impact of Randomization on the Efficiency of Neural Sequence Classification – We propose a method to identify the optimal number to sequence the training data in time for evaluating different models over different sets of data. We show that this method could outperform existing methods with respect to both accuracy and efficiency, especially when […]

Sequence Induction and Optimization for Embedding Storylets – The current work, based on the idea of the Kernelized Learning framework, is not only focused on the problems of prediction under noisy inputs but also to the problems of prediction under noisy inputs of the same name. A practical understanding of the problem of prediction under […]

Machine Learning Methods for Multi-Step Traffic Acquisition – Sparse-time classification (STR) has emerged as a promising tool for automatic vehicle identification. The main drawback of STR is its lack of training data and the difficulty of handling noisy data. In this work we present an innovative approach to the problem using Convolutional Neural Networks. In […]

Recurrent Neural Networks with Word-Partitioned LSTM for Action Recognition – This paper presents a novel method for learning to recognize human actions in a 3D environment using convolutional neural networks (CNN). Our first approach is a multi-level CNN trained with convolutional neural networks, where the CNN is given a low-level representation of the user object […]

Multi-label Visual Place Matching – A major challenge in the area of Convolutional Neural Networks (CNN) is the lack of explicit representation of multiple target regions. In this work, we present a novel method which enables the learning of multiple target regions without supervision (i.e., labeling) at each instant. The method is based on a […]

Probabilistic Latent Variable Models – In this paper, we present a new probabilistic model class, which is the same as classical logistic regression models and yet is better general. In previous work, we used Bayesian network and model parameters to model the problem of estimating the unknowns from the data. In this paper, we extend […]

Large-Scale Machine Learning for Classification – Many applications with a particular focus on a variety of complex datasets usually require very extensive training samples. In this paper, we focus on a class of data-driven classification problems, where it is challenging to accurately predict the classification results given a data set with a high-dimensional representation of […]

Efficient Bayesian Inference for Hidden Markov Models – We consider the problem of learning Markov auctions, where a user auctions an item and the auction proceeds according to some fixed value, where an auction value is generated by the user and a finite number of auctions are performed. Unlike the problem of auctions where the […]

Deep Reinforcement Learning based on Fuzzy IDP Recognition in Interactive Environments – In this work, we examine the effectiveness of deep neural networks for autonomous driving in scenarios involving high dynamic driving dynamics. Based on the recent advances in supervised learning and reinforcement learning, we devise a supervised learning process that produces novel driving behaviors […]