Our estimation methods enable practitioners to search for change points in settings where this was impossible before. The Loh-Wainwright based imputation method presented in Section 3.1 has a stable performance even in challenging high-dimensional s…

RASA nlu is a useful and well-evaluated toolkit for developing nlu components in sds and chatbot systems. We extended RASA by adding provisions for incremental processing generally, and we implemented two components for intent recognition that used …

We would like to thank Amin Aminzadeh Gohari, Salman Beigi, and Mohammad Hossein Yassaee for helpful discussions.

In this paper, we have presented the first provable RL algorithm with both polynomial runtime and polynomial sample complexity for linear MDPs, without requiring a “simulator” or additional assumptions. The algorithm is simply Least-Squares Value It…

In this section, we construct a low-variance estimate of the Lovász subgradient, and we apply the stochastic subgradient descent algorithm on it to minimize the Lovász extension. The stochastic subgradient descent method is a general algorithm for a…

In this paper, we proposed a novel monocular kinematic imaging system for assessing 3D postural sway during postural transition under varying cerebral perfusion levels. By physically embedding geometric priors, central and upper body kinematic motio…

In this work, we applied Gaussian Process classification to the problem of automated vehicle validation. It was found that the performance boundary, which separates successful from unsuccessful scenarios, can be adequately estimated from simulated d…

Time is an important feature in many applications involving events that occur synchronously and/or asynchronously. To effectively consume time information, recent studies have focused on designing new architectures. In this paper, we take an orthogo…

We propose an end-to-end data association model for online multi-object tracking. By jointly training the affinity learning module and the GNN optimization module, they can co-adapt collaboratively, improving the adaptivity, scalability, and accurac…

This paper provides a proof-of-concept that fully autonomous aerial animal biometrics is practically feasible. Operating in a real-world agricultural setting, the paper demonstrated that individual cattle identities can be reliably recovered biometr…

The structure of a Mixture Density Network, as laid out in detail in (?), allows us to sample our target predictions from a linear combination of m Gaussian distributions, each multiplied by an overall factor of αi, rather than from a single Gaussia…

In this paper, we considered the problem of deriving an approximation to kernel PCA using Nyström method. This latter approach seemingly overcomes some of the difficulties of other approaches based on random features. In particular, it allows to der…

Backed by our supplementary studies, we underpinned the large potential of animal avatars for VR research and applications, be it for education or entertainment. To provide a starting point for future research, we proposed a number of different cont…

We have introduced, retina-VAE, a novel application of variational autoencoders towards determining a clinically relevant latent space for exudative macular degeneration and related conditions which require treatment with anti-VEGF agents. The laten…

We proposed a cost-aware algorithm for interactive sequence-to-sequence learning, with a self-regulation module at its core that learns which type of feedback to query from a human teacher. The empirical study on interactive NMT with simulated human…

We investigated a machine-learning based approach to automated mechanism design and introduced the first truly general-purpose data representation, network architecture, and robust problem formulation to learn truthful and efficient auction rules au…

The basic proposed idea (possibly) can and needs to be extended in various ways: For instance, I have not discussed stochastic uncertainty: The data could be stochastic, and/or the evaluation of the objective may be stochastic.

Let us summarize our results. A strongly parallel algorithm MARS for the ground state searching problem for the fully connected Ising spin system is proposed. The performance of the algorithm is compared with the classical simulated annealing method…

Within this work, a novel, deep learning-based, method for the pixel-perfect detection and sizing of agglomerated, aggregated or occluded primary particles was proposed and validated. Therefore, the very powerful Mask R-CNNarchitecture was carefully…