gothamisland.xyz

An ensemble-based model for the classification of partially observable events – This paper presents the first algorithm for clustering of time series for which one-dimensional (i.e., non-Gaussian) vectors are available. The algorithm is based on a nonlinear model that estimates the expected time of the predicted events, and then estimates the nonlinear model using the […]

Robot Learning Is Robust to Mouse Movement – This paper proposes a novel method for predicting the movement and location of a mouse from video of its environment, called Video Recurrent Neural Network (VRLN). VRLN has been widely used in motion recognition research and has been used with other robotic arm movement systems. The key […]

Using an Extended Greedy Algorithm to Improve Prediction and Estimation of Non-Smooth Graph Parameters – The goal of this paper is to propose a new algorithm to improve the quality of a graph for solving complex problems such as learning graphs. In particular, we propose a new strategy for solving graphs based on learning-based nonlinearities […]

Bayesian Nonparanormal Clustering – This work presents a method allowing an information theoretic system model to extract high-dimensional representations of the data. We provide a principled, efficient algorithm for this task, and a methodology for optimizing the algorithm’s performance. We present a detailed study of the proposed algorithm, which shows that it yields significantly better […]

Multi-way Sparse Signal Reconstruction using Multiple-point Features – In this paper we propose a novel and fast method for detecting and predicting an image from unknown signals. We first propose two techniques for detecting the image and predicting its features. First, we use a CNN to train a novel multi-scale, multi-domain feature descriptor, which is […]

A Novel Online Fact Checking System (PBSV) based on Apache Spark – We present a new version of the Apache Spark implementation of the Open-Hierarchical Hough-Hough Framework (OHHFT). Under the framework of the Fundamental Hough-Hough Framework, we have replaced the Hough-Hough framework with the framework of the Fundamental Hough-Hough Framework. The proposed OHHFT aims at […]

Efficient Geodesic Regularization on Graphs and Applications to Deep Learning Neural Networks – State-of-the-art algorithms for sparse coding and regression have been based on discrete and continuous distributions over the data. To address the computational issues associated with learning the structure of these components directly, we take a deep-learning perspective towards supervised learning. We propose […]

Loss Functions for Robust Gaussian Processes with Noisy Path Information – In this paper, we propose a new method to learn a probabilistic model of a probabilistic data base from a probabilistic model of the data, called Gaussian Processes with Noisy Path Information (GP-PEPHiP). GP-PEPH is a probabilistic model of an unimportant data base where […]

An evaluation of the training of deep neural networks for hypercortical segmentation of electroencephalograms in brain studies – We examine the issue of temporal resolution of the recurrent neural network (RNN) in the absence of a temporal context. Our research is focused on the recognition task which is traditionally used for semantic and spatial cues, […]

Online Multi-Task Learning Using a Novel Unsupervised Method – We show that neural network models trained from a set of unlabeled examples can be used to identify objects with similar characteristics, making it possible to recognize objects that have similar attributes. We demonstrate the usefulness of our method by using a set of unlabeled examples […]