This research explores the integration of quantum data embedding techniques into classical machine learning (ML) algorithms, aiming to assess the performance enhancements and computational implications across a spectrum of models. We explore various classical-to-quantum mapping methods, ranging from basis encoding, angle encoding to amplitude encoding for encoding classical data, we conducted an extensive empirical study encompassing popular ML algorithms, including Logistic Regression, K-Nearest Neighbors, Support Vector Machines and ensemble methods like Random Forest, LightGBM, AdaBoost, and CatBoost. Our findings reveal that quantum data embedding contributes to improved classification accuracy and F1 scores, particularly notable in models that inherently benefit from enhanced feature representation. We observed nuanced effects on running time, with low-complexity models exhibiting moderate increases and more computationally intensive models experiencing discernible changes. Notably, ensemble methods demonstrated a favorable balance between performance gains and computational overhead. This study underscores the potential of quantum data embedding in enhancing classical ML models and emphasizes the importance of weighing performance improvements against computational costs. Future research directions may involve refining quantum encoding processes to optimize computational efficiency and exploring scalability for real-world applications. Our work contributes to the growing body of knowledge at the intersection of quantum computing and classical machine learning, offering insights for researchers and practitioners seeking to harness the advantages of quantum-inspired techniques in practical scenarios.

Audio adversarial examples (AEs) have posed significant security challenges to real-world speaker recognition systems. Most black-box attacks still require certain information from the speaker recognition model to be effective (e.g., keeping probing and requiring the knowledge of similarity scores). This work aims to push the practicality of the black-box attacks by minimizing the attacker's knowledge about a target speaker recognition model. Although it is not feasible for an attacker to succeed with completely zero knowledge, we assume that the attacker only knows a short (or a few seconds) speech sample of a target speaker. Without any probing to gain further knowledge about the target model, we propose a new mechanism, called parrot training, to generate AEs against the target model. Motivated by recent advancements in voice conversion (VC), we propose to use the one short sentence knowledge to generate more synthetic speech samples that sound like the target speaker, called parrot speech. Then, we use these parrot speech samples to train a parrot-trained(PT) surrogate model for the attacker. Under a joint transferability and perception framework, we investigate different ways to generate AEs on the PT model (called PT-AEs) to ensure the PT-AEs can be generated with high transferability to a black-box target model with good human perceptual quality. Real-world experiments show that the resultant PT-AEs achieve the attack success rates of 45.8% - 80.8% against the open-source models in the digital-line scenario and 47.9% - 58.3% against smart devices, including Apple HomePod (Siri), Amazon Echo, and Google Home, in the over-the-air scenario.

Recent large vision-language models such as CLIP have shown remarkable out-of-distribution (OOD) detection and generalization performance. However, their zero-shot in-distribution (ID) accuracy is often limited for downstream datasets. Recent CLIP-based fine-tuning methods such as prompt learning have demonstrated significant improvements in ID classification and OOD generalization where OOD labels are available. Nonetheless, it remains unclear whether the model is reliable to semantic shifts without OOD labels. In this paper, we aim to bridge the gap and present a comprehensive study to understand how fine-tuning impact OOD detection for few-shot downstream tasks. By framing OOD detection as multi-modal concept matching, we establish a connection between fine-tuning methods and various OOD scores. Our results suggest that a proper choice of OOD scores is essential for CLIP-based fine-tuning. In particular, the maximum concept matching (MCM) score provides a promising solution consistently. We also show that prompt learning demonstrates the state-of-the-art OOD detection performance over the zero-shot counterpart.

We develop a novel multiple hypothesis testing correction with family-wise error rate (FWER) control that efficiently exploits positive dependencies between potentially correlated statistical hypothesis tests. Our proposed algorithm $\texttt{max-rank}$ is conceptually straight-forward, relying on the use of a $\max$-operator in the rank domain of computed test statistics. We compare our approach to the frequently employed Bonferroni correction, theoretically and empirically demonstrating its superiority over Bonferroni in the case of existing positive dependency, and its equivalence otherwise. Our advantage over Bonferroni increases as the number of tests rises, and we maintain high statistical power whilst ensuring FWER control. We specifically frame our algorithm in the context of parallel permutation testing, a scenario that arises in our primary application of conformal prediction, a recently popularized approach for quantifying uncertainty in complex predictive settings.

The central question of ensemble learning has been: what factors make an ensemble system good or bad? It has We propose a fundamental theory on ensemble been widely believed that accurate and diverse models lead learning that answers the central question: what to better performance for ensemble systems. Guided by factors make an ensemble system good or bad? this intuition, many heuristical metrics have been proposed Previous studies used a variant of Fano's inequality to measure accuracy and diversity (Kohavi et al., 1996; of information theory and derived a lower Skalak et al., 1996; Cunningham & Carney, 2000; Shipp bound of the classification error rate on the basis & Kuncheva, 2002). However, these metrics lack theoretical of the accuracy and diversity of models. We grounding, and indeed, Kuncheva & Whitaker (2003) revisit the original Fano's inequality and argue empirically showed that there are no connections between that the studies did not take into account the information the metrics and system performance through a broad range lost when multiple model predictions of experiments. Turning to theoretical viewpoints, Geman are combined into a final prediction. To address et al. (1992) decomposed the squared error loss used in regression this issue, we generalize the previous theory to tasks into the bias and covariance of models. Bias incorporate the information loss, which we name here corresponds to accuracy and covariance diversity.

Ensuring a neural network is not relying on protected attributes (e.g., race, sex, age) for predictions is crucial in advancing fair and trustworthy AI. While several promising methods for removing attribute bias in neural networks have been proposed, their limitations remain under-explored. In this work, we mathematically and empirically reveal an important limitation of attribute bias removal methods in presence of strong bias. Specifically, we derive a general non-vacuous information-theoretical upper bound on the performance of any attribute bias removal method in terms of the bias strength. We provide extensive experiments on synthetic, image, and census datasets to verify the theoretical bound and its consequences in practice. Our findings show that existing attribute bias removal methods are effective only when the inherent bias in the dataset is relatively weak, thus cautioning against the use of these methods in smaller datasets where strong attribute bias can occur, and advocating the need for methods that can overcome this limitation.

We propose a semi-automatic staging area for efficiently building an accurate database of experimental physical properties of superconductors from literature, called SuperCon2, to enrich the existing manually-built superconductor database SuperCon. Here we report our curation interface (SuperCon2 Interface) and a workflow managing the state transitions of each examined record, to validate the dataset of superconductors from PDF documents collected using Grobid-superconductors in a previous work. This curation workflow allows both automatic and manual operations, the former contains ``anomaly detection'' that scans new data identifying outliers, and a ``training data collector'' mechanism that collects training data examples based on manual corrections. Such training data collection policy is effective in improving the machine-learning models with a reduced number of examples. For manual operations, the interface (SuperCon2 interface) is developed to increase efficiency during manual correction by providing a smart interface and an enhanced PDF document viewer. We show that our interface significantly improves the curation quality by boosting precision and recall as compared with the traditional ``manual correction''. Our semi-automatic approach would provide a solution for achieving a reliable database with text-data mining of scientific documents.

The performance of state-of-the-art machine learning models often deteriorates when testing on demographics that are under-represented in the training dataset. This problem has predominately been studied in a supervised learning setting where the data distribution is static. However, real-world applications often involve distribution shifts caused by the deployed models. For instance, the performance disparity against monitory users can lead to a high customer churn rate, thus the available data provided by active users are skewed due to the lack of minority users. This feedback effect further exacerbates the disparity among different demographic groups in future steps. To address this issue, we propose asymptotically fair participation as a condition to maintain long-term model performance over all demographic groups. In this work, we aim to address the problem of achieving asymptotically fair participation via optimal control formulation. Moreover, we design a surrogate retention system based on existing literature on evolutionary population dynamics to approximate the dynamics of distribution shifts on active user counts, from which the objective of achieving asymptotically fair participation is formulated as an optimal control problem, and the control variables are considered as the model parameters. We apply an efficient implementation of Pontryagin's maximum principle to estimate the optimal control solution. To evaluate the effectiveness of the proposed method, we design a generic simulation environment that simulates the population dynamics of the feedback effect between user retention and model performance. When we deploy the resulting models to the simulation environment, the optimal control solution accounts for long-term planning and leads to superior performance compared with existing baseline methods.

In the realm of scientific exploration and technological advancement, the use of Artificial Intelligence (AI) has catalyzed breakthroughs across various scientific and technological domains. One such domain that has witnessed significant transformation is materials science, where AI-driven approaches is believed to have the potential to revolutionize the search for novel materials with desired properties: towards this aim, data quality remains key in determining reliability of AI-models. Clearly, the quality of data is a multifaceted issue, as it is linked to disparate aspects in data generation including the accuracy by which materials properties are either measured or computed by simulations, the state of knowledge and/or ability to control operating parameters during experiments, the different adopted protocols and metrological approaches etc. In this work, we focus on a few special aspects of data quality that - to the best of our knowledge - have been poorly discussed in the literature, despite their possible detrimental impact on the ability of AI-based models to serve as platforms for material discovery. One of these aspects pertains to data bias, and while it has been previously mentioned in other works [1, 2], we believe that we still lack quantitative detection and assessment tools. We therefore present here a potential quantitative approach to assess it. Other overlooked aspects, namely possible hidden variables and disparate data age, are also discussed.

Amidst growing concerns of large language models (LLMs) being misused for generating misinformation or completing homework assignments, watermarking has emerged as an effective solution for distinguishing human-written and LLM-generated text. A prominent watermarking strategy is to embed a signal into generated text by upsampling a (pseudorandomly-chosen) subset of tokens at every generation step. Although this signal is imperceptible to a human reader, it is detectable through statistical testing. However, implanting such signals alters the model's output distribution and can have unintended effects when watermarked LLMs are used for downstream applications. In this work, we evaluate the performance of watermarked LLMs on a diverse suite of tasks, including text classification, textual entailment, reasoning, question answering, translation, summarization, and language modeling. We find that watermarking has negligible impact on the performance of tasks posed as k-class classification problems in the average case. However, the accuracy can plummet to that of a random classifier for some scenarios (that occur with non-negligible probability). Tasks that are cast as multiple-choice questions and short-form generation are surprisingly unaffected by watermarking. For long-form generation tasks, including summarization and translation, we see a drop of 15-20% in the performance due to watermarking. Our findings highlight the trade-offs that users should be cognizant of when using watermarked models, and point to cases where future research could improve existing trade-offs.