Estimating Energy Requirements for Computation of Complex Interactions – We present a method for inferring the relative costs of two types of interactions: (1) a dynamic equilibrium where the costs of two types interact and (2) a heterogeneous equilibrium where the costs of heterogeneous types interact jointly. Such a model can also be used to determine the relative value of the costs of different types of interactions. In the last research work, we show that the cost function of a dynamic equilibrium can be expressed as a finite-state program. This program can be learned with finite energy consumption. We provide a new method of inferring the relative costs of heterogeneous and dynamic equilibrium. We compare the two types of interactions and demonstrate how these two types of interactions can be learned.

This paper addresses the problem of extracting semantic features from textual data. We firstly present a new semantic segmentation method, namely Multistructure-Based Semantic Segmentation (MBSSE), that takes advantage of a semantic segmentation model to obtain better semantic features than the existing ones. Empirical evaluations on three datasets, including the MS-10 dataset, also demonstrate performance improvement over the existing ones. Furthermore, we compare MBSSE with a state-of-the-art semantic segmentation method, based on the Multistructure-based Temporal Segmentation.

A Robust Method for Non-Stationary Stochastic Regression

A Large Benchmark Dataset for Video Grounding and Tracking

Estimating Energy Requirements for Computation of Complex Interactions

Learning to Generate Chairs with Pointwise Loss Functions

Recognizing and Improving Textual Video by Interpreting Video DescriptionsThis paper addresses the problem of extracting semantic features from textual data. We firstly present a new semantic segmentation method, namely Multistructure-Based Semantic Segmentation (MBSSE), that takes advantage of a semantic segmentation model to obtain better semantic features than the existing ones. Empirical evaluations on three datasets, including the MS-10 dataset, also demonstrate performance improvement over the existing ones. Furthermore, we compare MBSSE with a state-of-the-art semantic segmentation method, based on the Multistructure-based Temporal Segmentation.