Search

Text-to-Speech (TTS) and Voice Conversion (VC) models have exhibited remarkable performance in generating realistic and natural audio. However, their dark side, audio deepfake poses a significant threat to both society and individuals. Existing countermeasures largely focus on determining the genuineness of speech based on complete original audio recordings, which however often contain private content. This oversight may refrain deepfake detection from many applications, particularly in scenarios involving sensitive information like business secrets. In this paper, we propose SafeEar, a novel framework that aims to detect deepfake audios without relying on accessing the speech content within. Our key idea is to devise a neural audio codec into a novel decoupling model that well separates the semantic and acoustic information from audio samples, and only use the acoustic information (e.g., prosody and timbre) for deepfake detection. In this way, no semantic content will be exposed to the detector. To overcome the challenge of identifying diverse deepfake audio without semantic clues, we enhance our deepfake detector with real-world codec augmentation. Extensive experiments conducted on four benchmark datasets demonstrate SafeEar's effectiveness in detecting various deepfake techniques with an equal error rate (EER) down to 2.02%. Simultaneously, it shields five-language speech content from being deciphered by both machine and human auditory analysis, demonstrated by word error rates (WERs) all above 93.93% and our user study. Furthermore, our benchmark constructed for anti-deepfake and anti-content recovery evaluation helps provide a basis for future research in the realms of audio privacy preservation and deepfake detection., Comment: Accepted by ACM CCS 2024. Please cite this paper as "Xinfeng Li, Kai Li, Yifan Zheng, Chen Yan, Xiaoyu Ji, Wenyuan Xu. SafeEar: Content Privacy-Preserving Audio Deepfake Detection. In Proceedings of ACM Conference on Computer and Communications Security (CCS), 2024."

Object detection can localize and identify objects in images, and it is extensively employed in critical multimedia applications such as security surveillance and autonomous driving. Despite the success of existing object detection models, they are often evaluated in ideal scenarios where captured images guarantee the accurate and complete representation of the detecting scenes. However, images captured by image sensors may be affected by different factors in real applications, including cyber-physical attacks. In particular, attackers can exploit hardware properties within the systems to inject electromagnetic interference so as to manipulate the images. Such attacks can cause noisy or incomplete information about the captured scene, leading to incorrect detection results, potentially granting attackers malicious control over critical functions of the systems. This paper presents a research work that comprehensively quantifies and analyzes the impacts of such attacks on state-of-the-art object detection models in practice. It also sheds light on the underlying reasons for the incorrect detection outcomes., Comment: 2024 IEEE International Conference on Multimedia and Expo (ICME), July 15 - July 19, 2024, Niagra Falls, Ontario, Canada

Pulsar timing arrays (PTAs), aimed at detecting gravitational waves (GWs) in the $1\sim 100$ nHz range, have recently made significant strides. Compelling evidence has emerged for a common spectrum signal spatially correlated among pulsars, following a Hellings-Downs (HD) pattern, which is crucial for detecting a gravitational-wave background (GWB). However, the HD curve is expected for discrete and non-interfering sources, which is unlikely to hold in realistic scenarios with potential interference among numerous GW sources, such as the supermassive black-hole binaries. Incorporating interference was previously expected to introduce an irreducible uncertainty (known as "cosmic variance") in discerning the HD correlation; however, our work reveals how this interference generates measurable frequency-dependent spatial correlations distinct from the HD curve. The spatial correlations for interfering sources (referred to as "ISC") still exhibit contributions in the quadrupole and higher orders, resembling the HD correlation and encoding the nature of GW radiations. We apply these novel correlations to search for a GWB in the NANOGrav 15-year data set. In an optimistic estimation, our findings show a Bayes factor of $33.7\pm 3.2$ comparing ISC to the HD correlation, and an improvement in optimal statistic signal-to-noise ratio from $4.9\pm 1.1$ for the HD correlation to $6.6\pm 1.7$ for the ISC, highlighting the significant enhancement in evidence for detecting a GWB through incorporating interference., Comment: 18 pages, 6 figures; minor revision and typo corrected

GWnext 2024 was a meeting held in the Kavli Institute for Astronomy and Astrophysics at Peking University in March $4^\text{th} - 8^\text{th}$, 2024. In the meeting researchers at different career stages -- with a particular focus on early career scientists -- working on the different aspects of gravitational wave (GW) astronomy gathered to discuss the current status as well as prospects of the field. The meeting was divided into three core sessions: Astrophysics, GW Theory, and Detection. Each session consisted of introductory talks and extended discussion sessions. Moreover, there was a poster session where students could present their results. In this paper, we summarize the results presented during the meeting and present the most important outcomes.

Text-to-image (T2I) models, such as Stable Diffusion, have exhibited remarkable performance in generating high-quality images from text descriptions in recent years. However, text-to-image models may be tricked into generating not-safe-for-work (NSFW) content, particularly in sexually explicit scenarios. Existing countermeasures mostly focus on filtering inappropriate inputs and outputs, or suppressing improper text embeddings, which can block sexually explicit content (e.g., naked) but may still be vulnerable to adversarial prompts -- inputs that appear innocent but are ill-intended. In this paper, we present SafeGen, a framework to mitigate sexual content generation by text-to-image models in a text-agnostic manner. The key idea is to eliminate explicit visual representations from the model regardless of the text input. In this way, the text-to-image model is resistant to adversarial prompts since such unsafe visual representations are obstructed from within. Extensive experiments conducted on four datasets and large-scale user studies demonstrate SafeGen's effectiveness in mitigating sexually explicit content generation while preserving the high-fidelity of benign images. SafeGen outperforms eight state-of-the-art baseline methods and achieves 99.4% sexual content removal performance. Furthermore, our constructed benchmark of adversarial prompts provides a basis for future development and evaluation of anti-NSFW-generation methods., Comment: Accepted by ACM CCS 2024. Please cite this paper as "Xinfeng Li, Yuchen Yang, Jiangyi Deng, Chen Yan, Yanjiao Chen, Xiaoyu Ji, Wenyuan Xu. SafeGen: Mitigating Sexually Explicit Content Generation in Text-to-Image Models. In Proceedings of ACM Conference on Computer and Communications Security (CCS), 2024."

We adopt an optimal-control framework for addressing the undiscounted infinite-horizon discrete-time restless $N$-armed bandit problem. Unlike most studies that rely on constructing policies based on the relaxed single-armed Markov Decision Process (MDP), we propose relaxing the entire bandit MDP as an optimal-control problem through the certainty equivalence control principle. Our main contribution is demonstrating that the reachability of an optimal stationary state within the optimal-control problem is a sufficient condition for the existence of an asymptotically optimal policy. Such a policy can be devised using an "align and steer" strategy. This reachability assumption is less stringent than any prior assumptions imposed on the arm-level MDP, notably the unichain condition is no longer needed. Through numerical examples, we show that employing model predictive control for steering generally results in superior performance compared to other existing policies., Comment: 10 pages, 3 figures, submitted to CDC 2024

The recent detection of a stochastic signal in the NANOGrav 15-year data set has aroused great interest in uncovering its origin. However, the evidence for the Hellings-Downs correlations, a key signature of the gravitational-wave background (GWB) predicted by general relativity, remains inconclusive. In this letter, we search for an isotropic non-tensorial GWB, allowed by general metric theories of gravity, in the NANOGrav 15-year data set. Our analysis reveals a Bayes factor of approximately 2.5, comparing the quadrupolar (tensor transverse, TT) correlations to the scalar transverse (ST) correlations, suggesting that the ST correlations provide a comparable explanation for the observed stochastic signal in the NANOGrav data. We obtain the median and the $90\%$ equal-tail amplitudes as $\mathcal{A}_\mathrm{ST} = 7.8^{+5.1}_{-3.5} \times 10^{-15}$ at the frequency of 1/year. Furthermore, we find that the vector longitudinal (VL) and scalar longitudinal (SL) correlations are weakly and strongly disfavoured by data, respectively, yielding upper limits on the amplitudes: $\mathcal{A}_\mathrm{VL}^{95\%} \lesssim 1.7 \times 10^{-15}$ and $\mathcal{A}_\mathrm{SL}^{95\%} \lesssim 7.4 \times 10^{-17}$. Lastly, we fit the NANOGrav data with the general transverse (GT) correlations parameterized by a free parameter $\alpha$. Our analysis yields $\alpha=1.74^{+1.18}_{-1.41}$, thus excluding both the TT ($\alpha=3$) and ST ($\alpha=0$) models at the $90\%$ confidence level., Comment: 9 pages, 4 figures, 3 tables; match the published version

General relativity predicts that gravitational waves propagate at the speed of light. Although ground-based gravitational-wave detectors have successfully constrained the velocity of gravitational waves in the high-frequency range, extending this constraint to the lower frequency range remains a challenge. In this work, we utilize the deviations in the overlap reduction function for a gravitational-wave background within pulsar timing arrays to investigate the velocity of gravitational waves in the nanohertz frequency band. By analyzing the NANOGrav 15-year data set, we obtain a well-constrained lower bound for the velocity of gravitational waves that $v \gtrsim 0.87\,c$, where $c$ is the speed of light., Comment: 6 pages, 1 figure, 1 table

The recently detected stochastic signal by several pulsar timing array collaborations, offers an opportunity to scrutinize the fundamental properties of gravity, including the potential mass of the graviton. In this study, we analyze the NANOGrav 15-year data set to search for a stochastic gravitational wave background with modified Hellings-Downs correlations predicted by massive gravity. While the Bayesian analysis comparing the massive gravity to massless gravity within the effective searchable mass range of $m_g\in [3\times 10^{-25}, 8 \times 10^{-24}]\,\rm{eV}/c^2$ does not yield an explicit upper bound as all the Bayes factors are smaller than $3$, the combined consideration of the minimum frequency inherent in a massive gravity and the observed spectrum leads to an upper limit of $m_g<8.2\times 10^{-24}\,\rm{eV}/c^2$., Comment: 6 pages, 1 figure, 1 table

In the frequency band of Laser Interferometer Space Antenna (LISA), extensive research has been conducted on the impact of foreground confusion noise generated by galactic binaries within the Milky Way galaxy. Additionally, the recent evidence for a stochastic signal, announced by the NANOGrav, EPTA, PPTA, CPTA and InPTA, indicates that the stochastic gravitational-wave background generated by supermassive black hole binaries (SMBHBs) can contribute a strong background noise within in LISA band. Given the presence of such strong noise, it is expected to have a considerable impacts on LISA's scientific missions. In this work, we investigate the impacts of the SGWB generated by SMBHBs on the detection of massive black hole binaries (MBHBs), verified galactic binaries (VGBs) and extreme mass ratio inspirals (EMRIs) in the context of LISA, and find it crucial to resolve and eliminate the exceed noise from the SGWB to ensure the success of LISA's missions., Comment: 6 pages, 3 figures

We examine a multi-stage stochastic optimization problem characterized by stagewise-independent, decision-dependent noises with strict constraints. The problem assumes convexity in that, following a specific relaxation, it transforms into a deterministic convex program. The relaxation process is inspired by the principle of Certainty Equivalence Control, which substitutes uncertainties with their nominal values and requires the hard constraints to be satisfied only in an expected sense. Utilizing the solutions obtained from these convex programs, we propose two universal methodologies -- re-solving-based and projection-based -- to formulate feasible policies relevant to the original problem. These methodologies are subsequently amalgamated to develop a hybrid policy, equipped with a tuning parameter that manages the frequency of re-solving. We derive upper bounds on the gap between the performance of these heuristic policies and the optimal one. Under the Lipschitz-type regularity of the model, these bounds are proportional to the square root of the stochastic noise variance. Assuming additional $C^2$-smoothness regularity, an alternative bound, proportional to the variance of the stochastic noise, can be established -- providing a refinement when variances are small. We also delve into the issues concerning the exponential growth of a multiplicative constant in these bounds with respect to the stage number. Our model strives to lay a comprehensive framework for dynamic sequential decision-making in the presence of uncertainties, incorporating classic problems from inventory management and Markovian bandits, while also accommodating a wider spectrum of stochastic optimization challenges. Specifically, we illustrate our methods on a network utility maximization problem through numerical experiments., Comment: 69 pages, 5 figures, submitted to Operations Research

Automatic speech recognition (ASR) systems have been shown to be vulnerable to adversarial examples (AEs). Recent success all assumes that users will not notice or disrupt the attack process despite the existence of music/noise-like sounds and spontaneous responses from voice assistants. Nonetheless, in practical user-present scenarios, user awareness may nullify existing attack attempts that launch unexpected sounds or ASR usage. In this paper, we seek to bridge the gap in existing research and extend the attack to user-present scenarios. We propose VRIFLE, an inaudible adversarial perturbation (IAP) attack via ultrasound delivery that can manipulate ASRs as a user speaks. The inherent differences between audible sounds and ultrasounds make IAP delivery face unprecedented challenges such as distortion, noise, and instability. In this regard, we design a novel ultrasonic transformation model to enhance the crafted perturbation to be physically effective and even survive long-distance delivery. We further enable VRIFLE's robustness by adopting a series of augmentation on user and real-world variations during the generation process. In this way, VRIFLE features an effective real-time manipulation of the ASR output from different distances and under any speech of users, with an alter-and-mute strategy that suppresses the impact of user disruption. Our extensive experiments in both digital and physical worlds verify VRIFLE's effectiveness under various configurations, robustness against six kinds of defenses, and universality in a targeted manner. We also show that VRIFLE can be delivered with a portable attack device and even everyday-life loudspeakers., Comment: Accepted by NDSS Symposium 2024. Please cite this paper as "Xinfeng Li, Chen Yan, Xuancun Lu, Zihan Zeng, Xiaoyu Ji, Wenyuan Xu. Inaudible Adversarial Perturbation: Manipulating the Recognition of User Speech in Real Time. In Network and Distributed System Security (NDSS) Symposium 2024."

NANOGrav, EPTA, PPTA, and CPTA have announced the evidence for a stochastic signal from their latest data sets. Supermassive black hole binaries (SMBHBs) are supposed to be the most promising gravitational-wave (GW) sources of pulsar timing arrays. Assuming an astro-informed formation model, we use the NANOGrav 15-year data set to constrain the gravitational wave background (GWB) from SMBHBs. Our results prefer a large turn-over eccentricity of the SMBHB orbit when GWs begin to dominate the SMBHBs evolution. Furthermore, the GWB spectrum is extrapolated to the space-borne GW detector frequency band by including inspiral-merge-cutoff phases of SMBHBs and should be detected by LISA, Taiji and TianQin in the near future., Comment: 8 pages, 3 figures, 3 tables; version accepted for publication in SCPMA

Automatic Speaker Recognition Systems (SRSs) have been widely used in voice applications for personal identification and access control. A typical SRS consists of three stages, i.e., training, enrollment, and recognition. Previous work has revealed that SRSs can be bypassed by backdoor attacks at the training stage or by adversarial example attacks at the recognition stage. In this paper, we propose Tuner, a new type of backdoor attack against the enrollment stage of SRS via adversarial ultrasound modulation, which is inaudible, synchronization-free, content-independent, and black-box. Our key idea is to first inject the backdoor into the SRS with modulated ultrasound when a legitimate user initiates the enrollment, and afterward, the polluted SRS will grant access to both the legitimate user and the adversary with high confidence. Our attack faces a major challenge of unpredictable user articulation at the enrollment stage. To overcome this challenge, we generate the ultrasonic backdoor by augmenting the optimization process with random speech content, vocalizing time, and volume of the user. Furthermore, to achieve real-world robustness, we improve the ultrasonic signal over traditional methods using sparse frequency points, pre-compensation, and single-sideband (SSB) modulation. We extensively evaluate Tuner on two common datasets and seven representative SRS models, as well as its robustness against seven kinds of defenses. Results show that our attack can successfully bypass speaker recognition systems while remaining effective to various speakers, speech content, etc. To mitigate this newly discovered threat, we also provide discussions on potential countermeasures, limitations, and future works of this new threat., Comment: Published in Internet of Things Journal (IoT-J)

An important goal in artificial intelligence is to create agents that can both interact naturally with humans and learn from their feedback. Here we demonstrate how to use reinforcement learning from human feedback (RLHF) to improve upon simulated, embodied agents trained to a base level of competency with imitation learning. First, we collected data of humans interacting with agents in a simulated 3D world. We then asked annotators to record moments where they believed that agents either progressed toward or regressed from their human-instructed goal. Using this annotation data we leveraged a novel method - which we call "Inter-temporal Bradley-Terry" (IBT) modelling - to build a reward model that captures human judgments. Agents trained to optimise rewards delivered from IBT reward models improved with respect to all of our metrics, including subsequent human judgment during live interactions with agents. Altogether our results demonstrate how one can successfully leverage human judgments to improve agent behaviour, allowing us to use reinforcement learning in complex, embodied domains without programmatic reward functions. Videos of agent behaviour may be found at https://youtu.be/v_Z9F2_eKk4.

We propose a new policy, called the LP-update policy, to solve finite horizon weakly-coupled Markov decision processes. The latter can be seen as multi-constraint multi-action bandits, and generalize the classical restless bandit problems. Our solution is based on re-solving periodically a relaxed version of the original problem, that can be cast as a linear program (LP). When the problem is made of $N$ statistically identical sub-components, we show that the LP-update policy becomes asymptotically optimal at rate $O(T^2/\sqrt{N})$. This rate can be improved to $O(T/\sqrt{N})$ if the problem satisfies some ergodicity property and to $O(1/N)$ if the problem is non-degenerate. The definition of non-degeneracy extends the same notion for restless bandits. By using this property, we also improve the computational efficiency of the LP-update policy. We illustrate the performance of our policy on randomly generated examples, as well as a generalized applicant screening problem, and show that it outperforms existing heuristics., Comment: 29 pages. Preprint

Human language learners are exposed to a trickle of informative, context-sensitive language, but a flood of raw sensory data. Through both social language use and internal processes of rehearsal and practice, language learners are able to build high-level, semantic representations that explain their perceptions. Here, we take inspiration from such processes of "inner speech" in humans (Vygotsky, 1934) to better understand the role of intra-agent speech in embodied behavior. First, we formally pose intra-agent speech as a semi-supervised problem and develop two algorithms that enable visually grounded captioning with little labeled language data. We then experimentally compute scaling curves over different amounts of labeled data and compare the data efficiency against a supervised learning baseline. Finally, we incorporate intra-agent speech into an embodied, mobile manipulator agent operating in a 3D virtual world, and show that with as few as 150 additional image captions, intra-agent speech endows the agent with the ability to manipulate and answer questions about a new object without any related task-directed experience (zero-shot). Taken together, our experiments suggest that modelling intra-agent speech is effective in enabling embodied agents to learn new tasks efficiently and without direct interaction experience.

Creating agents that can interact naturally with humans is a common goal in artificial intelligence (AI) research. However, evaluating these interactions is challenging: collecting online human-agent interactions is slow and expensive, yet faster proxy metrics often do not correlate well with interactive evaluation. In this paper, we assess the merits of these existing evaluation metrics and present a novel approach to evaluation called the Standardised Test Suite (STS). The STS uses behavioural scenarios mined from real human interaction data. Agents see replayed scenario context, receive an instruction, and are then given control to complete the interaction offline. These agent continuations are recorded and sent to human annotators to mark as success or failure, and agents are ranked according to the proportion of continuations in which they succeed. The resulting STS is fast, controlled, interpretable, and representative of naturalistic interactions. Altogether, the STS consolidates much of what is desirable across many of our standard evaluation metrics, allowing us to accelerate research progress towards producing agents that can interact naturally with humans. A video may be found at https://youtu.be/YR1TngGORGQ.

Using mathematical modeling and human subjects experiments, this research explores the extent to which emerging webcams might leak recognizable textual and graphical information gleaming from eyeglass reflections captured by webcams. The primary goal of our work is to measure, compute, and predict the factors, limits, and thresholds of recognizability as webcam technology evolves in the future. Our work explores and characterizes the viable threat models based on optical attacks using multi-frame super resolution techniques on sequences of video frames. Our models and experimental results in a controlled lab setting show it is possible to reconstruct and recognize with over 75% accuracy on-screen texts that have heights as small as 10 mm with a 720p webcam. We further apply this threat model to web textual contents with varying attacker capabilities to find thresholds at which text becomes recognizable. Our user study with 20 participants suggests present-day 720p webcams are sufficient for adversaries to reconstruct textual content on big-font websites. Our models further show that the evolution towards 4K cameras will tip the threshold of text leakage to reconstruction of most header texts on popular websites. Besides textual targets, a case study on recognizing a closed-world dataset of Alexa top 100 websites with 720p webcams shows a maximum recognition accuracy of 94% with 10 participants even without using machine-learning models. Our research proposes near-term mitigations including a software prototype that users can use to blur the eyeglass areas of their video streams. For possible long-term defenses, we advocate an individual reflection testing procedure to assess threats under various settings, and justify the importance of following the principle of least privilege for privacy-sensitive scenarios.

We investigate the evolution of phase transition of the classical fully compact dimer model on the bipartite square lattice with second nearest bonds at finite temperatures. We use the numeric Monte Carlo method with the directed-loop algorithm to simulate the model. Our results show that the order of the phase transition depends on the fugacity of the second nearest bonds. We find that the phase transition reduces from the Kosterlitz-Thouless transition to unconventional high-order phase transitions which feature the coexistence of properties Kosterliz-Thouless transition and the first-order phases transition simultaneously. As the fugacity increases further, phase transition evolves to the first-order phase transition. In addition, our results of dimer-dimer correlation functions and their corresponding structure factor functions computed by us show the evolution of decay correlation for different fugacity.

Traffic light recognition is essential for fully autonomous driving in urban areas. In this paper, we investigate the feasibility of fooling traffic light recognition mechanisms by shedding laser interference on the camera. By exploiting the rolling shutter of CMOS sensors, we manage to inject a color stripe overlapped on the traffic light in the image, which can cause a red light to be recognized as a green light or vice versa. To increase the success rate, we design an optimization method to search for effective laser parameters based on empirical models of laser interference. Our evaluation in emulated and real-world setups on 2 state-of-the-art recognition systems and 5 cameras reports a maximum success rate of 30% and 86.25% for Red-to-Green and Green-to-Red attacks. We observe that the attack is effective in continuous frames from more than 40 meters away against a moving vehicle, which may cause end-to-end impacts on self-driving such as running a red light or emergency stop. To mitigate the threat, we propose redesigning the rolling shutter mechanism., Comment: To be published in USENIX Security 2022

A common vision from science fiction is that robots will one day inhabit our physical spaces, sense the world as we do, assist our physical labours, and communicate with us through natural language. Here we study how to design artificial agents that can interact naturally with humans using the simplification of a virtual environment. We show that imitation learning of human-human interactions in a simulated world, in conjunction with self-supervised learning, is sufficient to produce a multimodal interactive agent, which we call MIA, that successfully interacts with non-adversarial humans 75% of the time. We further identify architectural and algorithmic techniques that improve performance, such as hierarchical action selection. Altogether, our results demonstrate that imitation of multi-modal, real-time human behaviour may provide a straightforward and surprisingly effective means of imbuing agents with a rich behavioural prior from which agents might then be fine-tuned for specific purposes, thus laying a foundation for training capable agents for interactive robots or digital assistants. A video of MIA's behaviour may be found at https://youtu.be/ZFgRhviF7mY

Inferring the abstract relational and causal structure of the world is a major challenge for reinforcement-learning (RL) agents. For humans, language--particularly in the form of explanations--plays a considerable role in overcoming this challenge. Here, we show that language can play a similar role for deep RL agents in complex environments. While agents typically struggle to acquire relational and causal knowledge, augmenting their experience by training them to predict language descriptions and explanations can overcome these limitations. We show that language can help agents learn challenging relational tasks, and examine which aspects of language contribute to its benefits. We then show that explanations can help agents to infer not only relational but also causal structure. Language can shape the way that agents to generalize out-of-distribution from ambiguous, causally-confounded training, and explanations even allow agents to learn to perform experimental interventions to identify causal relationships. Our results suggest that language description and explanation may be powerful tools for improving agent learning and generalization., Comment: ICML 2022; 23 pages

We provide a framework to analyse control policies for the restless Markovian bandit model, under both finite and infinite time horizon. We show that when the population of arms goes to infinity, the value of the optimal control policy converges to the solution of a linear program (LP). We provide necessary and sufficient conditions for a generic control policy to be: i) asymptotically optimal; ii) asymptotically optimal with square root convergence rate; iii) asymptotically optimal with exponential rate. We then construct the LP-index policy that is asymptotically optimal with square root convergence rate on all models, and with exponential rate if the model is non-degenerate in finite horizon, and satisfies a uniform global attractor property in infinite horizon. We next define the LP-update policy, which is essentially a repeated LP-index policy that solves a new linear program at each decision epoch. We provide numerical experiments to compare the efficiency of LP-based policies. We compare the performance of the LP-index policy and the LP-update policy with other heuristics. Our result demonstrates that the LP-update policy outperforms the LP-index policy in general, and can have a significant advantage when the transition matrices are wrongly estimated., Comment: Mathematics of Operations Research, 2023

We evaluate the performance of Whittle index policy for restless Markovian bandits, when the number of bandits grows. It is proven in [30] that this performance is asymptotically optimal if the bandits are indexable and the associated deterministic system has a global attractor fixed point. In this paper we show that, under the same conditions, the convergence rate is exponential in the number of bandits, unless the fixed point is singular (to be defined later). Our proof is based on the nature of the deterministic equation governing the stochastic system: We show that it is a piecewise affine continuous dynamical system inside the simplex of the empirical measure of the bandits. Using simulations and numerical solvers, we also investigate the cases where the conditions for the exponential rate theorem are violated, notably when attracting limit cycles appear, or when the fixed point is singular. We illustrate our theorem on a Markovian fading channel model, which has been well studied in the literature. Finally, we extend our synchronous model results to the asynchronous model.

A common vision from science fiction is that robots will one day inhabit our physical spaces, sense the world as we do, assist our physical labours, and communicate with us through natural language. Here we study how to design artificial agents that can interact naturally with humans using the simplification of a virtual environment. This setting nevertheless integrates a number of the central challenges of artificial intelligence (AI) research: complex visual perception and goal-directed physical control, grounded language comprehension and production, and multi-agent social interaction. To build agents that can robustly interact with humans, we would ideally train them while they interact with humans. However, this is presently impractical. Therefore, we approximate the role of the human with another learned agent, and use ideas from inverse reinforcement learning to reduce the disparities between human-human and agent-agent interactive behaviour. Rigorously evaluating our agents poses a great challenge, so we develop a variety of behavioural tests, including evaluation by humans who watch videos of agents or interact directly with them. These evaluations convincingly demonstrate that interactive training and auxiliary losses improve agent behaviour beyond what is achieved by supervised learning of actions alone. Further, we demonstrate that agent capabilities generalise beyond literal experiences in the dataset. Finally, we train evaluation models whose ratings of agents agree well with human judgement, thus permitting the evaluation of new agent models without additional effort. Taken together, our results in this virtual environment provide evidence that large-scale human behavioural imitation is a promising tool to create intelligent, interactive agents, and the challenge of reliably evaluating such agents is possible to surmount.

With the both spin and valley degrees of freedom, the low-lying excitonic spectra of photo-excited transition-metal dichalcogenide monolayers (TMDC-MLs) are featured by rich fine structures, comprising the intra-valley bright exciton states as well as various intra- and inter-valley dark ones. The latter states can be classified as those of the spin- and momentum-forbidden dark excitons according to the violated optical selection rules. Because of the optical invisibility, the two types of the dark states are in general hardly observed and even distinguished in conventional spectroscopies although their impacts on the optical and dynamical properties of TMDC-MLs have been well noticed. In this Letter, we present a theoretical and computational investigation of the exciton fine structures and the temperature-dependent photo-luminescence spectra of strained tungsten diselenide monolayers (WSe$_2$-MLs) where the intra-valley spin-forbidden dark exciton lies in the lowest exciton states and other momentum-forbidden states are in the higher energies that are tunable by external stress. The numerical computations are carried out by solving the Bethe-Salpeter equation for an exciton in a WSe$_2$-ML under the stress-control in the tight-binding scheme established from the first principle computation in the density functional theory. According to the numerical computation and supportive model analysis, we reveal the distinctive signatures of the spin- and momentum-forbidden exciton states of strained WSe$_2$-MLs in the temperature-dependent photo-luminescences and present the guiding principle to infer the relative energetic locations of the two types of DX's., Comment: 44 pages, 7 figures

In this work, we investigate the generation of second harmonic light by gold nanorods and demonstrate that the collected nonlinear intensity depends upon a phase interplay between different modes available in the nanostructure. By recording the backward and forward emitted second harmonic signals from nanorods with various lengths, we find that the maximum nonlinear signal emitted in the forward and backward directions is not obtained for the same nanorod length. We confirm the experimental results with the help of full-wave computations done with a surface integral equation method. These observations are explained by the multipolar nature of the second harmonic emission, which emphasizes the role played by the relative phase between the second harmonic modes. Our findings are of a particular importance for the design of plasmonic nanostructures with controllable nonlinear emission and nonlinear plasmonic sensors as well as for the coherent control of harmonic generations in plasmonic nanostructures.

Speech recognition (SR) systems such as Siri or Google Now have become an increasingly popular human-computer interaction method, and have turned various systems into voice controllable systems(VCS). Prior work on attacking VCS shows that the hidden voice commands that are incomprehensible to people can control the systems. Hidden voice commands, though hidden, are nonetheless audible. In this work, we design a completely inaudible attack, DolphinAttack, that modulates voice commands on ultrasonic carriers (e.g., f > 20 kHz) to achieve inaudibility. By leveraging the nonlinearity of the microphone circuits, the modulated low frequency audio commands can be successfully demodulated, recovered, and more importantly interpreted by the speech recognition systems. We validate DolphinAttack on popular speech recognition systems, including Siri, Google Now, Samsung S Voice, Huawei HiVoice, Cortana and Alexa. By injecting a sequence of inaudible voice commands, we show a few proof-of-concept attacks, which include activating Siri to initiate a FaceTime call on iPhone, activating Google Now to switch the phone to the airplane mode, and even manipulating the navigation system in an Audi automobile. We propose hardware and software defense solutions. We validate that it is feasible to detect DolphinAttack by classifying the audios using supported vector machine (SVM), and suggest to re-design voice controllable systems to be resilient to inaudible voice command attacks., Comment: 15 pages, 17 figures

A biconcave particle suspended in a Poiseuille flow is investigated by the multiple-relaxation-time lattice Boltzmann method with the Galilean-invariant momentum exchange method. The lateral migration and equilibrium of the particle are similar to the Segr\'e-Silberberg effect in our numerical simulations. Surprisingly, two lateral equilibrium positions are observed corresponding to the releasing positions of the biconcave particle. The upper equilibrium positions significantly decrease with the growth of the Reynolds number, whereas the lower ones are almost insensitive to the Reynolds number. Interestingly, the regular wave accompanied by nonuniform rotation is exhibited in the lateral movement of the biconcave particle. It can be attributed to that the biconcave shape in various postures interacts with the parabolic velocity distribution of the Poiseuille flow. A set of contours illustrate the dynamic flow field when the biconcave particle has successive postures in a rotating period., Comment: 13 pages, 5 figures

An artificial neural network (ANN) algorithm, employing several Aqua MODIS infrared channels, the retrieved total cloud visible optical depth, and vertical humidity profiles is trained to detect multilayer (ML) ice-over-water cloud systems as identified by matched CloudSat and CALIPSO (CC) data. The multilayer ANN, or MLANN, algorithm is also trained to retrieve the optical depth and the top and base heights of the upper-layer ice clouds in ML systems. The trained MLANN was applied to independent MODIS data resulting in a combined ML and single layer hit rate of 80% (77%) for nonpolar regions during the day (night). The results are more accurate than currently available methods and the previous version of the MLANN. Upper-layer cloud top and base heights are accurate to ±1.2 km and ±1.6 km, respectively, while the uncertainty in optical depth is ±0.457 and ±0.556 during day and night, respectively. Areas of further improvement and development are identified and will be addressed in future versions of the MLANN.

The Clouds and Earth's Radiant Energy System (CERES) Project is producing a suite of cloud properties from high-resolution imagers on several satellites and matching them precisely with broadband radiance data to study the influence of clouds and radiation on climate. The cloud properties generally compare well with independent validation sources. Distinct differences are found between the CERES cloud properties and those derived with other algorithms from the same imager data. CERES products will be updated beginning in late 2006.