Search

Knowledge transfer in multi-task learning is typically viewed as a dichotomy; positive transfer, which improves the performance of all tasks, or negative transfer, which hinders the performance of all tasks. In this paper, we investigate the understudied problem of asymmetric task relationships, where knowledge transfer aids the learning of certain tasks while hindering the learning of others. We propose an optimisation strategy that includes additional cloned tasks named self-auxiliaries into the learning process to flexibly transfer knowledge between tasks asymmetrically. Our method can exploit asymmetric task relationships, benefiting from the positive transfer component while avoiding the negative transfer component. We demonstrate that asymmetric knowledge transfer provides substantial improvements in performance compared to existing multi-task optimisation strategies on benchmark computer vision problems.

Having a detailed knowledge of the intracluster medium (ICM) to infer the exact cluster physics such as the cluster dynamical state is crucial for cluster-based cosmological studies. This knowledge limits the accuracy and precision of mass estimation, a key parameter for such studies. In this paper, we conduct an in-depth analysis of cluster ACT-CL J0240.0+0116 using a multi-wavelength approach, with a primary focus on high angular resolution Sunyaev-Zeldovich (SZ) thermal component observations obtained under the NIKA2 Sunyaev-Zeldovich Large Programme (LPSZ). We create composite images using NIKA2, X-ray, and optical galaxy number density maps. The results reveal distinct signs of disturbance within the cluster with the distributions of gas and member galaxies that do not overlap. We also find suggestions of an inflow of matter onto the cluster from the southwestern direction. Ultimately, we classify the cluster as disturbed, using morphological indicators derived from its SZ, X-ray, and optical image. The cluster SZ signal is also contaminated by a strong central point source. We adopt different approaches to handling this contaminant and find the estimates of our pressure and hydrostatic mass profiles robust to the point source mitigation model. The cluster hydrostatic mass is estimated at $4.25^{+0.50}_{-0.45\, } \times 10^{14} \,\mathrm{M}_{\odot}$ for the case where the point source was masked. These values are consistent with the mass estimated using only X-ray data and with those from previous SZ studies of the Atacama cosmology telescope (ACT) survey, with improved precision on the mass estimate. Our findings strongly suggest that ACT-CL J0240.0+0116 is a disturbed cluster system, and the detailed observations and derived values serve as a compelling case study for the capabilities of the LPSZ in mapping the cluster ICM with high precision., Comment: 13 pages, 7 figures, submitted to A&A

Differential privacy upper-bounds the information leakage of machine learning models, yet providing meaningful privacy guarantees has proven to be challenging in practice. The private prediction setting where model outputs are privatized is being investigated as an alternate way to provide formal guarantees at prediction time. Most current private prediction algorithms, however, rely on global sensitivity for noise calibration, which often results in large amounts of noise being added to the predictions. Data-specific noise calibration, such as smooth sensitivity, could significantly reduce the amount of noise added, but were so far infeasible to compute exactly for modern machine learning models. In this work we provide a novel and practical approach based on convex relaxation and bound propagation to compute a provable upper-bound for the local and smooth sensitivity of a prediction. This bound allows us to reduce the magnitude of noise added or improve privacy accounting in the private prediction setting. We validate our framework on datasets from financial services, medical image classification, and natural language processing and across models and find our approach to reduce the noise added by up to order of magnitude., Comment: 16 pages, 11 figures

We demonstrate nearly a microsecond of spin coherence in Er3+ ions doped in cerium dioxide nanocrystal hosts, despite a large gyromagnetic ratio and nanometric proximity of the spin defect to the nanocrystal surface. The long spin coherence is enabled by reducing the dopant density below the instantaneous diffusion limit in a nuclear spin-free host material, reaching the limit of a single erbium spin defect per nanocrystal. We observe a large Orbach energy in a highly symmetric cubic site, further protecting the coherence in a qubit that would otherwise rapidly decohere. Spatially correlated electron spectroscopy measurements reveal the presence of Ce3+ at the nanocrystal surface that likely acts as extraneous paramagnetic spin noise. Even with these factors, defect-embedded nanocrystal hosts show tremendous promise for quantum sensing and quantum communication applications, with multiple avenues, including core-shell fabrication, redox tuning of oxygen vacancies, and organic surfactant modification, available to further enhance their spin coherence and functionality in the future., Comment: 26 pages, 5 figures

Modern machine learning pipelines leverage large amounts of public data, making it infeasible to guarantee data quality and leaving models open to poisoning and backdoor attacks. Provably bounding model behavior under such attacks remains an open problem. In this work, we address this challenge by developing the first framework providing provable guarantees on the behavior of models trained with potentially manipulated data without modifying the model or learning algorithm. In particular, our framework certifies robustness against untargeted and targeted poisoning, as well as backdoor attacks, for bounded and unbounded manipulations of the training inputs and labels. Our method leverages convex relaxations to over-approximate the set of all possible parameter updates for a given poisoning threat model, allowing us to bound the set of all reachable parameters for any gradient-based learning algorithm. Given this set of parameters, we provide bounds on worst-case behavior, including model performance and backdoor success rate. We demonstrate our approach on multiple real-world datasets from applications including energy consumption, medical imaging, and autonomous driving., Comment: 21 pages, 8 figures

Optically interfaced solid-state defects are promising candidates for quantum communication technologies. The ideal defect system would feature bright telecom emission, long-lived spin states, and a scalable material platform, simultaneously. Here, we employ one such system, vanadium (V4+) in silicon carbide (SiC), to establish a potential telecom spin-photon interface within a mature semiconductor host. This demonstration of efficient optical spin polarization and readout facilitates all optical measurements of temperature-dependent spin relaxation times (T1). With this technique, we lower the temperature from about 2K to 100 mK to observe a remarkable four-orders-of-magnitude increase in spin T1 from all measured sites, with site-specific values ranging from 57 ms to above 27 s. Furthermore, we identify the underlying relaxation mechanisms, which involve a two-phonon Orbach process, indicating the opportunity for strain-tuning to enable qubit operation at higher temperatures. These results position V4+ in SiC as a prime candidate for scalable quantum nodes in future quantum networks., Comment: 11 pages, 6 figures

We detail several scientific and engineering innovations which enable the controlled spalling of 10 - 50 micron thick films of single crystal 4H silicon carbide (4H-SiC) from bulk substrates. 4H-SiC's properties, including high thermal conductivity and a wide bandgap, make it an ideal candidate for high-temperature, high-voltage power electronic devices. Moreover, 4H-SiC has been shown to be an excellent host of solid-state atomic defect qubits for quantum computing and quantum networking. Because 4H-SiC single crystal substrates are expensive (due to long growth times and limited yield), techniques for removal and transfer of bulk-quality films in the tens-of-microns thickness range are highly desirable to allow for substrate reuse and integration of the separated films. In this work we utilize novel approaches for stressor layer thickness control and spalling crack initiation to demonstrate controlled spalling of 4H-SiC, the highest fracture toughness material spalled to date. Additionally, we demonstrate substrate re-use, bonding of the spalled films to carrier substrates, and explore the spin coherence of the spalled films. In preliminary studies we are able to achieve coherent spin control of neutral divacancy ($VV^{0}$) qubit ensembles and measure a quasi-bulk spin $T_{2}$ of 79.7 $\mu$s in such spalled films., Comment: 16 pages, 6 figures

Identifying black holes is essential for comprehending the development of stars and uncovering novel principles of physics. Gravitational microlensing provides an exceptional opportunity to examine an undetectable population of black holes in the Milky Way. In particular, long-lasting events are likely to be associated with massive lenses, including black holes. We present an analysis of the Gaia18ajz microlensing event, reported by the Gaia Science Alerts system, which has exhibited a long timescale and features indicative of the annual microlensing parallax effect. Our objective is to estimate the parameters of the lens based on the best-fitting model. We utilized photometric data obtained from the Gaia satellite and terrestrial observatories to investigate a variety of microlensing models and calculate the most probable mass and distance to the lens, taking into consideration a Galactic model as a prior. Subsequently, weapplied a mass-brightness relation to evaluate the likelihood that the lens is a main sequence star. We also describe the DarkLensCode (DLC), an open-source routine which computes the distribution of probable lens mass, distance and luminosity employing the Galaxy priors on stellar density and velocity for microlensing events with detected microlensing parallax. We modelled Gaia18ajz event and found its two possible models with most likely Einstein timescale of $316^{+36}_{-30}$ days and $299^{+25}_{-22}$ days. Applying Galaxy priors for stellar density and motion, we calculated the most probable lens mass of $4.9^{+5.4}_{-2.3} M_\odot$ located at $1.14^{+0.75}_{-0.57}\,\text{kpc}$ or $11.1^{+10.3}_{-4.7} M_\odot$ located at $1.31^{+0.80}_{-0.60}\,\text{kpc}$. Our analysis of the blended light suggests that the lens is likely a dark remnant of stellar evolution, rather than a main sequence star.

With the advancement in robotics, it is becoming increasingly common for large factories and warehouses to incorporate visual SLAM (vSLAM) enabled automated robots that operate closely next to humans. This makes any adversarial attacks on vSLAM components potentially detrimental to humans working alongside them. Loop Closure Detection (LCD) is a crucial component in vSLAM that minimizes the accumulation of drift in mapping, since even a small drift can accumulate into a significant drift over time. A prior work by Kim et al., SymbioLCD2, unified visual features and semantic objects into a single graph structure for finding loop closure candidates. While this provided a performance improvement over visual feature-based LCD, it also created a single point of vulnerability for potential graph-based adversarial attacks. Unlike previously reported visual-patch based attacks, small graph perturbations are far more challenging to detect, making them a more significant threat. In this paper, we present Adversarial-LCD, a novel black-box evasion attack framework that employs an eigencentrality-based perturbation method and an SVM-RBF surrogate model with a Weisfeiler-Lehman feature extractor for attacking graph-based LCD. Our evaluation shows that the attack performance of Adversarial-LCD with the SVM-RBF surrogate model was superior to that of other machine learning surrogate algorithms, including SVM-linear, SVM-polynomial, and Bayesian classifier, demonstrating the effectiveness of our attack framework. Furthermore, we show that our eigencentrality-based perturbation method outperforms other algorithms, such as Random-walk and Shortest-path, highlighting the efficiency of Adversarial-LCD's perturbation selection method., Comment: Accepted at VISIGRAPP 2024, 8 pages

Providing formal guarantees of algorithmic fairness is of paramount importance to socially responsible deployment of machine learning algorithms. In this work, we study formal guarantees, i.e., certificates, for individual fairness (IF) of neural networks. We start by introducing a novel convex approximation of IF constraints that exponentially decreases the computational cost of providing formal guarantees of local individual fairness. We highlight that prior methods are constrained by their focus on global IF certification and can therefore only scale to models with a few dozen hidden neurons, thus limiting their practical impact. We propose to certify distributional individual fairness which ensures that for a given empirical distribution and all distributions within a $\gamma$-Wasserstein ball, the neural network has guaranteed individually fair predictions. Leveraging developments in quasi-convex optimization, we provide novel and efficient certified bounds on distributional individual fairness and show that our method allows us to certify and regularize neural networks that are several orders of magnitude larger than those considered by prior works. Moreover, we study real-world distribution shifts and find our bounds to be a scalable, practical, and sound source of IF guarantees., Comment: 21 Pages, Neural Information Processing Systems 2023

Aspect-based sentiment analysis (ABSA) is a fine-grained type of sentiment analysis that identifies aspects and their associated opinions from a given text. With the surge of digital opinionated text data, ABSA gained increasing popularity for its ability to mine more detailed and targeted insights. Many review papers on ABSA subtasks and solution methodologies exist, however, few focus on trends over time or systemic issues relating to research application domains, datasets, and solution approaches. To fill the gap, this paper presents a systematic literature review (SLR) of ABSA studies with a focus on trends and high-level relationships among these fundamental components. This review is one of the largest SLRs on ABSA. To our knowledge, it is also the first to systematically examine the interrelations among ABSA research and data distribution across domains, as well as trends in solution paradigms and approaches. Our sample includes 727 primary studies screened from 8550 search results without time constraints via an innovative automatic filtering process. Our quantitative analysis not only identifies trends in nearly two decades of ABSA research development but also unveils a systemic lack of dataset and domain diversity as well as domain mismatch that may hinder the development of future ABSA research. We discuss these findings and their implications and propose suggestions for future research.

The performance of machine learning models depends on the quality of the underlying data. Malicious actors can attack the model by poisoning the training data. Current detectors are tied to either specific data types, models, or attacks, and therefore have limited applicability in real-world scenarios. This paper presents a novel fully-agnostic framework, DIVA (Detecting InVisible Attacks), that detects attacks solely relying on analyzing the potentially poisoned data set. DIVA is based on the idea that poisoning attacks can be detected by comparing the classifier's accuracy on poisoned and clean data and pre-trains a meta-learner using Complexity Measures to estimate the otherwise unknown accuracy on a hypothetical clean dataset. The framework applies to generic poisoning attacks. For evaluation purposes, in this paper, we test DIVA on label-flipping attacks., Comment: 8 pages

Machine learning models are increasingly used in fields that require high reliability such as cybersecurity. However, these models remain vulnerable to various attacks, among which the adversarial label-flipping attack poses significant threats. In label-flipping attacks, the adversary maliciously flips a portion of training labels to compromise the machine learning model. This paper raises significant concerns as these attacks can camouflage a highly skewed dataset as an easily solvable classification problem, often misleading machine learning practitioners into lower defenses and miscalculations of potential risks. This concern amplifies in tabular data settings, where identifying true labels requires expertise, allowing malicious label-flipping attacks to easily slip under the radar. To demonstrate this risk is inherited in the adversary's objective, we propose FALFA (Fast Adversarial Label-Flipping Attack), a novel efficient attack for crafting adversarial labels. FALFA is based on transforming the adversary's objective and employs linear programming to reduce computational complexity. Using ten real-world tabular datasets, we demonstrate FALFA's superior attack potential, highlighting the need for robust defenses against such threats., Comment: 10 pages

Model-based reinforcement learning seeks to simultaneously learn the dynamics of an unknown stochastic environment and synthesise an optimal policy for acting in it. Ensuring the safety and robustness of sequential decisions made through a policy in such an environment is a key challenge for policies intended for safety-critical scenarios. In this work, we investigate two complementary problems: first, computing reach-avoid probabilities for iterative predictions made with dynamical models, with dynamics described by Bayesian neural network (BNN); second, synthesising control policies that are optimal with respect to a given reach-avoid specification (reaching a "target" state, while avoiding a set of "unsafe" states) and a learned BNN model. Our solution leverages interval propagation and backward recursion techniques to compute lower bounds for the probability that a policy's sequence of actions leads to satisfying the reach-avoid specification. Such computed lower bounds provide safety certification for the given policy and BNN model. We then introduce control synthesis algorithms to derive policies maximizing said lower bounds on the safety probability. We demonstrate the effectiveness of our method on a series of control benchmarks characterized by learned BNN dynamics models. On our most challenging benchmark, compared to purely data-driven policies the optimal synthesis algorithm is able to provide more than a four-fold increase in the number of certifiable states and more than a three-fold increase in the average guaranteed reach-avoid probability., Comment: 47 pages, 10 figures. arXiv admin note: text overlap with arXiv:2105.10134

We study the problem of certifying the robustness of Bayesian neural networks (BNNs) to adversarial input perturbations. Given a compact set of input points $T \subseteq \mathbb{R}^m$ and a set of output points $S \subseteq \mathbb{R}^n$, we define two notions of robustness for BNNs in an adversarial setting: probabilistic robustness and decision robustness. Probabilistic robustness is the probability that for all points in $T$ the output of a BNN sampled from the posterior is in $S$. On the other hand, decision robustness considers the optimal decision of a BNN and checks if for all points in $T$ the optimal decision of the BNN for a given loss function lies within the output set $S$. Although exact computation of these robustness properties is challenging due to the probabilistic and non-convex nature of BNNs, we present a unified computational framework for efficiently and formally bounding them. Our approach is based on weight interval sampling, integration, and bound propagation techniques, and can be applied to BNNs with a large number of parameters, and independently of the (approximate) inference method employed to train the BNN. We evaluate the effectiveness of our methods on various regression and classification tasks, including an industrial regression benchmark, MNIST, traffic sign recognition, and airborne collision avoidance, and demonstrate that our approach enables certification of robustness and uncertainty of BNN predictions.

We study Individual Fairness (IF) for Bayesian neural networks (BNNs). Specifically, we consider the $\epsilon$-$\delta$-individual fairness notion, which requires that, for any pair of input points that are $\epsilon$-similar according to a given similarity metrics, the output of the BNN is within a given tolerance $\delta>0.$ We leverage bounds on statistical sampling over the input space and the relationship between adversarial robustness and individual fairness to derive a framework for the systematic estimation of $\epsilon$-$\delta$-IF, designing Fair-FGSM and Fair-PGD as global,fairness-aware extensions to gradient-based attacks for BNNs. We empirically study IF of a variety of approximately inferred BNNs with different architectures on fairness benchmarks, and compare against deterministic models learnt using frequentist techniques. Interestingly, we find that BNNs trained by means of approximate Bayesian inference consistently tend to be markedly more individually fair than their deterministic counterparts.

Running complex sets of machine learning experiments is challenging and time-consuming due to the lack of a unified framework. This leaves researchers forced to spend time implementing necessary features such as parallelization, caching, and checkpointing themselves instead of focussing on their project. To simplify the process, in this paper, we introduce Memento, a Python package that is designed to aid researchers and data scientists in the efficient management and execution of computationally intensive experiments. Memento has the capacity to streamline any experimental pipeline by providing a straightforward configuration matrix and the ability to concurrently run experiments across multiple threads. A demonstration of Memento is available at: https://wickerlab.org/publication/memento.

We analysed spectroscopic data obtained with VLT-VIMOS for two multiple-cluster systems, PLCKG$214.6+36.9$ and PLCKG$334.8-38.0$, discovered via their thermal Sunyaev-Zel'dovich signal by $Planck$. Combining the Optical spectroscopy, for the redshift determination, and photometric data from galaxy surveys (SDSS, WISE, DESI), we were able to study the structure of the two multiple-cluster systems, to determine their nature and the properties of their member galaxies. We found that the two systems are populated mainly with passive galaxies and that PLCKG$214.6+36.9$ consists of a pair of clusters at redshift $z = 0.445$ and a background isolated cluster at $z = 0.498$, whereas the system PLCKG$334.8-38.0$ is a chance association of three independent clusters at redshifts $z = 0.367$, $z =0.292$, and $z = 0.33$. We also find evidence for remaining star formation activity in the highest-redshift cluster of PLCKG$214.6+36.9$, at $z = 0.498$., Comment: 12 pages, 9 Figures, 5 Tables. Submitted to A&A, comments are welcome

Machine learning from explanations (MLX) is an approach to learning that uses human-provided explanations of relevant or irrelevant features for each input to ensure that model predictions are right for the right reasons. Existing MLX approaches rely on local model interpretation methods and require strong model smoothing to align model and human explanations, leading to sub-optimal performance. We recast MLX as a robustness problem, where human explanations specify a lower dimensional manifold from which perturbations can be drawn, and show both theoretically and empirically how this approach alleviates the need for strong model smoothing. We consider various approaches to achieving robustness, leading to improved performance over prior MLX methods. Finally, we show how to combine robustness with an earlier MLX method, yielding state-of-the-art results on both synthetic and real-world benchmarks., Comment: NeurIPS 2023; https://github.com/vihari/robust_mlx

Various white-dwarf (WD) binary scenarios have been proposed trying to understand the nature and the diversity of Type Ia supernovae (SNe Ia). In this work, we study the evolution of carbon-oxygen WD -- red giant (RG) binaries (including the role of magnetic confinement) as possible SN Ia progenitors (the so-called symbiotic progenitor channel). Using the \textsc{mesa} stellar evolution code, we calculate the time dependence of the structure of the RG star, the wind mass loss, the Roche-lobe-overflow (RLOF) mass-transfer rate, the polar mass-accretion rate (in the case of magnetic confinement), and the orbital and angular-momentum evolution. We consider cases where the WD is non-magnetic and cases where the magnetic field is strong enough to force accretion onto the two small polar caps of the WD. Confined accretion onto a small area allows for more efficient hydrogen burning, potentially suppressing nova outbursts. This makes it easier for the WD to grow in mass towards the Chandrasekhar mass limit and explode as a SN Ia. With magnetic confinement, the initial parameter space of the symbiotic channel for SNe Ia is shifted towards shorter orbital periods and lower donor masses compared to the case without magnetic confinement. Searches for low-mass He WDs or relatively low-mass giants with partially stripped envelopes that survived the supernova explosion and are found in SN remnants will provide crucial insights for our understanding of the contribution of this symbiotic channel., Comment: Published in The Astrophysical Journal Letters (published 2022 December 22)

Post-hoc explanation methods are used with the intent of providing insights about neural networks and are sometimes said to help engender trust in their outputs. However, popular explanations methods have been found to be fragile to minor perturbations of input features or model parameters. Relying on constraint relaxation techniques from non-convex optimization, we develop a method that upper-bounds the largest change an adversary can make to a gradient-based explanation via bounded manipulation of either the input features or model parameters. By propagating a compact input or parameter set as symbolic intervals through the forwards and backwards computations of the neural network we can formally certify the robustness of gradient-based explanations. Our bounds are differentiable, hence we can incorporate provable explanation robustness into neural network training. Empirically, our method surpasses the robustness provided by previous heuristic approaches. We find that our training method is the only method able to learn neural networks with certificates of explanation robustness across all six datasets tested., Comment: 23 pages, 12 figures