In this work, we proposed subset sampling and online hyper-parameter selection to speed up and enhance PNNL algorithms. Empirical results demonstrated with PMLP show that proposed approach can not only accelerate the optimization procedure in PMLP b…

In this paper, we propose a novel provably efficient recursion-based algorithm for agnostic Q-learning with general function approximation with bounded Eluder dimension in deterministic systems. We obtain a sharp characterization on the relation bet…

Non-line-of-sight (NLOS) imaging is a rapidly growing field seeking to form images of objects outside the field of view, with potential applications in search and rescue, reconnaissance, and even medical imaging. The critical challenge of NLOS imagi…

We addressed the problem of annotation gap existence in input sensor readings of ADL systems. We investigated three paradigms for handling these gaps. The first paradigm drops the non-annotated gaps from the training data. The second paradigm gives …

We have shown that elastic weight consolidation (EWC), a commonly-used regularization-based continual learning approach, is vulnerable to backdoor poisoning attacks. This vulnerability exists due to the way that EWC functions in retaining informatio…

In this work we present a general framework to adapt neural sequence models (Transformer) for conditional computation and control the amount of computation used at inference. Our proposed approach injects simple control networks into the core comput…

In this work, we propose the Augmented Normalizing Flows and a corresponding variational lower bound on the marginal likelihood. We show that the proposed method can be used to approximate a Hamiltonian dynamical system as a universal transport map …

In this paper, we have proposed a hybrid method to use the voxelized 3D+t NURBS-based surfaces of the XCAT heart model in a deep generative network and synthesize semantically and anatomically meaningful 4D realistic CMR images with controllable gro…

We developed Survival-OPT, a gradient-free algorithm that can create physical attacks in the hard-label black-box setting. We introduce the notion of survivability, which we can utilize in GFO optimization. We show that Survival-OPT is efficient: it…

In this paper we introduced a new method based on CNN to automatically choose RF frames that are suitable for strain estimation. Our method is fast, practical and does not need any external hardware. Therefore, it could be used commercially to gener…

Next, we study the generalization ability of GNNs via Rademacher bounds, focusing on binary classification. We generalize the previous results on the complexity of feedforward networks (BFT2017; NBS2018) and RNNs (CLZ2019) in mainly three ways. Firs…