Abstract

In safety-critical RL settings, the inclusion of an additional cost function is often favoured over the arduous task of modifying the reward function to ensure the agent's safe behaviour. However, designing or evaluating such a cost function can be prohibitively expensive. For instance, in the domain of self-driving, designing a cost function that encompasses all unsafe behaviours (eg, aggressive lane changes, risky overtakes) is inherently complex, it must also consider all the actors present in the scene making it expensive to evaluate. In such scenarios, the cost function can be learned from feedback collected offline in between training rounds. This feedback can be system generated or elicited from a human observing the training process. Previous approaches have not been able to scale to complex environments and are constrained to receiving feedback at the state level which can be expensive to collect. To this end, we introduce an approach that scales to more complex domains and extends beyond state-level feedback, thus, reducing the burden on the evaluator. Inferring the cost function in such settings poses challenges, particularly in assigning credit to individual states based on trajectory-level feedback. To address this, we propose a surrogate objective that transforms the problem into a state-level supervised classification task with noisy labels, which can be solved efficiently. Additionally, it is often infeasible to collect feedback for every trajectory generated by the agent, hence, two fundamental questions arise:(1) Which trajectories should be presented to the human? and (2) How many trajectories are necessary for effective learning?

Abstract

Preserving the privacy of preferences (or rewards) of a sequential decision-making agent when decisions are observable is crucial in many physical and cybersecurity domains. For instance, in wildlife monitoring, forest rangers must conduct surveillance without revealing animal locations to poachers. This paper addresses privacy preservation in planning over a sequence of actions in MDPs, where the reward function represents the preference structure to be protected. Observers can use Inverse RL (IRL) to learn these preferences, making this a challenging task. Current research on Differential Privacy (DP) in this setting fails to ensure a lower bound on the minimum expected reward and offers theoretical guarantees that are inadequate against IRL-based observers. To bridge this gap, we propose a novel approach rooted in the theory of deception. Deception includes two models: dissimulation (hiding the truth) and simulation (showing the wrong). As our first contribution, we theoretically demonstrate a significant privacy leak in the current dissimulation-based method. Our second contribution is a novel RL-based planning algorithm that uses simulation to effectively address these privacy concerns while ensuring a guarantee on the expected reward. Through experimentation on multiple benchmark problems, we show that our proposed approach outperforms existing methods in preserving the privacy of reward functions.

Abstract

Bell non-locality and steering are archetypal characteristics of quantum mechanics that mark a significant departure from conventional classical notions. They basically refer to the presence of quantum correlations between separated systems which violate a non-local inequality, the violation otherwise is not possible if we restrict ourselves only to classical correlations. In view of the importance of such unique correlations, one may be interested to generate more states exhibiting non-locality starting from a few, a protocol which is termed as broadcasting. However, in the present submission, we show using universal Buzek–Hillary (BH) quantum cloning machine that, if one restricts to broadcasting through local quantum cloning, then such non-locality cannot be broadcasted. Our study is done in the purview of the Bell-CHSH inequality and the CJWR [E G Cavalcanti et al, Phys. Rev. A 80, 032112 (2009)] steering inequality. It is observed that when the number of measurement settings is greater than 6, some of the output states are steerable after the application of local optimal BH cloners. We found that, under some restrictions, if the Werner and Bell diagonal states are subjected to such procedures, then the resultant state is rendered unsteerable. We extended this study to three qubit systems and found that genuine tripartite non-locality cannot be broadcasted using universal BH local quantum cloners, when the Svetlichny’s inequality was considered. For two qubit systems, we considered simulated general local unitaries over Werner and Bell-diagonal states and found that for none of these states, broadcasting on non-locality and 3-steerability is possible.

Abstract

Free-chlorine concentration monitoring is of importance in public and industrial water supplies. Current colorimetric methods, which include test strips, spectrophotometric kits, etc. either lack precision or are expensive and labor intensive. In this study, we present a fully automated, cost-effective method of measurement of free chlorine concentration in real -time. The setup includes an automatic powder dispenser, an automatic liquid dispenser, a sample chamber, and an LED-light-dependent resistor sensor pair. The liquid sample is mixed with a coloring reagent and its color is measured using the sensor pair. Different regression algorithms were trained on the sensor data and tuned to predict the corresponding free-chlorine concentration with maximum accuracy. The system eliminates the need for color matching, reduces the time taken per test, and can be used to predict concentrations of multiple analytes, including ammonia-nitrogen, dissolved oxygen, etc., by adding corresponding colorimetry agents. This allows for a fully automated, real-time water testing system.

Abstract

Analysis of foot pressure, also known as plantar pressure analysis, plays a pivotal role in biomedical assessments related to posture and gait analysis. Extensive research has been conducted on leveraging this technique for clinical purposes, leading to the development of flexible pressure sensors. In this letter, we present the use of in-shoe flexible pressure sensor array for determining the nature of the walking surface. The sensor system is fabricated using eight low-cost and robust, in-shoe pressure sensors that leverage the piezoresistivity of velostat. The sensor array was characterized for four different surface types. Random Forest (RF) algorithm was used to classify the surfaces with 86% accuracy. Based on this analysis, we propose a novel method for analyzing various surfaces based on their attributes such as firmness, rigidity, and penetrability.

Abstract

The rapid proliferation of Internet of Things (IoT) devices and applications has resulted in an increasing demand for Low Power and Wide Area Network (LPWAN) solutions. The adoption of IoT networks still faces several challenges, despite the rapid advancement of low-power communication technology. Homogenizing this sector requires allowing interoperability between many technologies, which is now one of the largest obstacles. In this article, we present the design and implementation of the hybrid LPWAN architecture that can accomplish wide-area communication coverage and low-power consumption for IoT applications by leveraging two LPWAN technologies, Wireless Smart Ubiquitous Network (Wi-SUN) and Long Range (LoRa). In particular, LoRa is used for long-range communication, and Wi-SUN for a low-latency mesh network. Additionally, we implemented smart street light controlling system as a real-world deployment at the university campus to showcase the efficiency of the hybrid network. Our results demonstrate that the hybrid LPWAN architecture provides a better coverage and capacity while consuming less power than that of the LoRa or Wi-SUN network. The results of this study demonstrate the effectiveness of the proposed hybrid LPWAN architecture as a viable solution for next-generation IoT applications.

Abstract

Affordable healthcare, particularly for elderly persons, is an important goal with aging of the world population. Cloud-connected smart sensor systems are an important part of the healthcare ecosystem. In this letter, we present an affordable large area flexible pressure sensor array using a piezoresistive thin film. The dimensions of the sensor are 180 cm × 90 cm, according to the dimensions of a standard single bed mattress. The sensor array consists of eight vertical sensor strips with a high aspect ratio of 120, highest reported so far for a flexible piezoresistive sensor. The measured output of the sensors showed a high signal-to-noise ratio, with loaded and unloaded sensor resistances of 11 pm 3 Omega and 219.2 pm 0.5 kOmega, respectively. The fabrication process has been designed to be conducive to roll-to-roll manufacturing processes to increase scalability and reduce cost. With a sampling frequency of 10 Hz, the sensor array provides real-time information about user position and possible fall events.

Abstract

We investigate the phenomenon of transition to synchronization in the Sakaguchi-Kuramoto model in the presence of higher-order interactions and global order parameter adaptation. The investigation is done by performing extensive numerical simulations and low-dimensional modeling of the system. Numerical simulations of the full system show both continuous (second-order) as well as discontinuous transitions. The discontinuous transitions can either be associated with explosive (first-order) or tiered synchronization states depending on the choice of parameters. To develop an in depth understanding of the transition scenario in the parameter space we derive a reduced order model (ROM) using the Ott-Antonsen ansatz, the results of which closely match with those of the numerical simulations of the full system. The simplicity and analytical accessibility of the ROM help to conveniently unfold the transition scenario in the system having complex dependence on the parameters. Simultaneous analysis of the full system and the ROM clearly identifies the regions of the parameter space exhibiting different types of transitions. It is observed that the second-order transition is connected with a supercritical pitchfork bifurcation (PB) of the ROM. On the other hand, the discontinuous tiered transition is associated with multiple saddle-node (SN) bifurcations along with a supercritical PB and the first-order transition involves a subcritical PB alongside a SN bifurcation. Finally, the stability analysis of the different synchronization states of the system is performed analytically.

Abstract

This paper identifies the cryptocurrency market crashes and analyses its dynamics using the complex network. We identify three distinct crashes during 2017–20, and the analysis is carried out by dividing the time series into pre-crash, crash, and post-crash periods. Partial correlation based complex network analysis is carried out to study the crashes. Degree density (𝜌𝐷), average path length (̄𝑙), and average clustering coefficient (𝑐𝑐) are estimated from these networks. We find that both 𝜌𝐷 and 𝑐𝑐 are smallest during the pre-crash period, and spike during the crash suggesting the network is dense during a crash. Although 𝜌𝐷 and 𝑐𝑐 decrease in the postcrash period, they remain higher than pre-crash levels for the 2017–18 and 2018–19 crashes suggesting a market attempt to return to normalcy. We get ̄𝑙 is minimal during the crash period, suggesting a rapid flow of information. A dense network and rapid information flow suggest that during a crash uninformed synchronized panic sell-off happens. However, during the 2019– 20 crash, the values of 𝜌𝐷, 𝑐𝑐, and ̄𝑙 did not vary significantly, indicating minimal change in dynamics compared to other crashes. The findings of this study may guide investors in making decisions during market crashes.

Abstract

Our investigation centres on assessing the importance of a biased parameter (𝛼) in a multiplex Markov chain (MMC) model that is characterized by biased random walks in multiplex networks. We explore how varying complex network topologies affect the total multiplex imbalance as a function of biased parameter. Our primary finding is that the system demonstrates a gradual increase in total imbalance within both positive and negative regions of the biased parameter, with a consistent minimum value occurring at 𝛼 = −1. In contrast to the negative region, the total imbalance is consistently high when 𝛼 is significantly positive. We perform a detailed examination of four different network structures and establish three sets of multiplex networks. In each of these networks, the second layer consists of a Regular Random network, while the first layer is either a Barabási–Albert, Erdős-Rényi, or Watts Strogatz network, depending on the set. Our results demonstrate that the combination of Barabási–Albert and Random Regular Network exhibits the highest level of right saturation imbalance. Additionally, for left saturation imbalance, the Erdős–Rényi and Random Regular combination achieve a slightly higher value. We also observe that the total amount of imbalance at 𝛼 = −1 follows a decreasing trend as the size of the network of each layer increases. Furthermore, we are also able to illustrate that the second most significant eigenvalue of the supra-transition matrix exhibits a similar pattern in response to changes in the bias parameter, aligning with the overall system’s imbalance.