Search for: All records

We study the problem of developing autonomous agents that can follow human instructions to infer and perform a sequence of actions to complete the underlying task. Significant progress has been made in recent years, especially for tasks with short horizons. However, when it comes to long-horizon tasks with extended sequences of actions, an agent can easily ignore some instructions or get stuck in the middle of the long instructions and eventually fail the task. To address this challenge, we propose a model-agnostic milestone-based task tracker(M-TRACK) to guide the agent and monitor its progress. Specifcally, we propose a milestone builder that tags the instructions with navigation and interaction milestones which the agent needs to complete step by step, and a milestone checker that systemically checks the agent’s progress in its current milestone and determines when to proceed to the next. On the challenging ALFRED dataset, our M-TRACK leads to a notable 33% and 52% relative improvement in unseen success rate over two competitive base models. 

We consider a group of agents that estimate their locations in an environment through sensor measurements and aim to transmit a message signal to a client via collaborative beamforming. Assuming that the localization error of each agent follows a Gaussian distribution, we study the problem of forming a reliable communication link between the agents and the client that achieves a desired signal-to-noise ratio (SNR) at the client with minimum variability. In particular, we develop a greedy subset selection algorithm that chooses only a subset of the agents to transmit the signal so that the variance of the received SNR is minimized while the expected SNR exceeds a desired threshold. We show the optimality of the proposed algorithm when the agents’ localization errors satisfy certain sufficient conditions that are characterized in terms of the carrier frequency. 

A quickest change detection problem is considered in a sensor network with observations whose statistical dependency structure across the sensors before and after the change is described by a decomposable graphical model (DGM). Distributed computation methods for this problem are proposed that are capable of producing the optimum centralized test statistic. The DGM leads to the proper way to collect nodes into local groups equivalent to cliques in the graph, such that a clique statistic which summarizes all the clique sensor data can be computed within each clique. The clique statistics are transmitted to a decision maker to produce the optimum centralized test statistic. In order to further improve communication efficiency, an ordered transmission approach is proposed where transmissions of the clique statistics to the fusion center are ordered and then adaptively halted when sufficient information is accumulated. This procedure is always guaranteed to provide the optimal change detection performance, despite not transmitting all the statistics from all the cliques. A lower bound on the average number of transmissions saved by ordered transmissions is provided and for the case where the change seldom occurs the lower bound approaches approximately half the number of cliques provided a well behaved distance measure between the distributions of the sensor observations before and after the change is sufficiently large. We also extend the approach to the case when the graph structure is different under each hypothesis. Numerical results show significant savings using the ordered transmission approach and validate the theoretical findings. 

Quickest change detection in a sensor network is considered where each sensor observes a sequence of random variables and transmits its local information on the observations to a fusion center. At an unknown point in time, the distribution of the observations at all sensors changes. The objective is to detect the change in distribution as soon as possible, subject to a false alarm constraint. We consider minimax formulations for this problem and propose a new approach where transmissions are ordered and halted when sufficient information is accumulated at the fusion center. We show that the proposed approach can achieve the optimal performance equivalent to the centralized cumulative sum (CUSUM) algorithm while requiring fewer sensor transmissions. Numerical results for a shift in mean of independent and identically distributed Gaussian observations show significant communication savings for the case where the change seldom occurs which is frequently true in many important applications. 

The topic of training machine learning models by employing multiple gradient-computing workers is attracting great interest recently. Communication efficiency in such distributed learning settings is an important consideration, especially for the case where the needed communications are expensive in terms of power usage. We develop a new approach which is efficient in terms of communication transmissions. In this scheme, only the most informative worker results are transmitted to reduce the total number of transmissions. Our ordered gradient approach provably achieves the same order of convergence rate as gradient descent for nonconvex smooth loss functions while gradient descent always requires more communications. Experiments show significant communication savings compared to the best existing approaches in some cases. 

Mining entity synonym sets (i.e., sets of terms referring to the same entity) is an important task for many entity-leveraging applications. Previous work either rank terms based on their similarity to a given query term, or treats the problem as a two-phase task (i.e., detecting synonymy pairs, followed by organizing these pairs into synonym sets). However, these approaches fail to model the holistic semantics of a set and suffer from the error propagation issue. Here we propose a new framework, named SynSetMine, that efficiently generates entity synonym sets from a given vocabulary, using example sets from external knowledge bases as distant supervision. SynSetMine consists of two novel modules: (1) a set-instance classifier that jointly learns how to represent a permutation invariant synonym set and whether to include a new instance (i.e., a term) into the set, and (2) a set generation algorithm that enumerates the vocabulary only once and applies the learned set-instance classifier to detect all entity synonym sets in it. Experiments on three real datasets from different domains demonstrate both effectiveness and efficiency of SynSetMine for mining entity synonym sets. 

One of the difficulties of implementing and analyzing algorithms that achieve information theoretic limits is adapting asymptotic results to the finite block-length regime. Results on secrecy for both regimes utilize Shannon entropy and mutual information as metrics for security. In this paper, we determine that Shannon entropy does not necessarily have equal utility for wireless authentication in finite block-length regimes with a focus on the fingerprint embedding framework. Then, we apply a new security performance metric to the framework that is linked to min-entropy rather than Shannon entropy and is similar to cheating probability used in the literature. The metric is based upon an adversary's ability to correctly guess the secret key over many observations using maximum likelihood decoding. We demonstrate the effect that system parameters such as the length of the key and the identification tag have on an adversary's ability to attack successfully. We find that if given a large key, it is better to use it all at once, than to use some and then renew the key with the remaining bits after a certain number of transmissions.