Search

Topic modeling is a fundamental task in natural language processing, allowing the discovery of latent thematic structures in text corpora. While Large Language Models (LLMs) have demonstrated promising capabilities in topic discovery, their direct application to topic modeling suffers from issues such as incomplete topic coverage, misalignment of topics, and inefficiency. To address these limitations, we propose LLM-ITL, a novel LLM-in-the-loop framework that integrates LLMs with many existing Neural Topic Models (NTMs). In LLM-ITL, global topics and document representations are learned through the NTM, while an LLM refines the topics via a confidence-weighted Optimal Transport (OT)-based alignment objective. This process enhances the interpretability and coherence of the learned topics, while maintaining the efficiency of NTMs. Extensive experiments demonstrate that LLM-ITL can help NTMs significantly improve their topic interpretability while maintaining the quality of document representation.

We introduce a novel state-space architecture for diffusion models, effectively harnessing spatial and frequency information to enhance the inductive bias towards local features in input images for image generation tasks. While state-space networks, including Mamba, a revolutionary advancement in recurrent neural networks, typically scan input sequences from left to right, they face difficulties in designing effective scanning strategies, especially in the processing of image data. Our method demonstrates that integrating wavelet transformation into Mamba enhances the local structure awareness of visual inputs and better captures long-range relations of frequencies by disentangling them into wavelet subbands, representing both low- and high-frequency components. These wavelet-based outputs are then processed and seamlessly fused with the original Mamba outputs through a cross-attention fusion layer, combining both spatial and frequency information to optimize the order awareness of state-space models which is essential for the details and overall quality of image generation. Besides, we introduce a globally-shared transformer to supercharge the performance of Mamba, harnessing its exceptional power to capture global relationships. Through extensive experiments on standard benchmarks, our method demonstrates superior results compared to DiT and DIFFUSSM, achieving faster training convergence and delivering high-quality outputs. The codes and pretrained models are released at https://github.com/VinAIResearch/DiMSUM.git., Comment: Accepted to NeurIPS 2024. Project page: https://hao-pt.github.io/dimsum/

Recently published work on rephrasing natural text data for pre-training LLMs has shown promising results when combining the original dataset with the synthetically rephrased data. We build upon previous work by replicating existing results on C4 and extending them with our optimized rephrasing pipeline to the English, German, Italian, and Spanish Oscar subsets of CulturaX. Our pipeline leads to increased performance on standard evaluation benchmarks in both the mono- and multilingual setup. In addition, we provide a detailed study of our pipeline, investigating the choice of the base dataset and LLM for the rephrasing, as well as the relationship between the model size and the performance after pre-training. By exploring data with different perceived quality levels, we show that gains decrease with higher quality. Furthermore, we find the difference in performance between model families to be bigger than between different model sizes. This highlights the necessity for detailed tests before choosing an LLM to rephrase large amounts of data. Moreover, we investigate the effect of pre-training with synthetic data on supervised fine-tuning. Here, we find increasing but inconclusive results that highly depend on the used benchmark. These results (again) highlight the need for better benchmarking setups. In summary, we show that rephrasing multilingual and low-quality data is a very promising direction to extend LLM pre-training data., Comment: 21 pages, 4 figures, 12 tables

We introduce the class of strongly sofic monoids. This class of monoids strictly contains the class of sofic groups and is a proper subclass of the class of sofic monoids. We define and investigate sofic topological entropy for actions of strongly sofic monoids on compact spaces. We show that sofic topological entropy is a topological conjugacy invariant for such actions and use this fact to prove that every strongly sofic monoid is surjunctive. This means that if $M$ is a strongly sofic monoid and $A$ is a finite alphabet set, then every injective cellular automaton $\tau \colon A^M \to A^M$ is surjective. As an application, we prove that the monoid algebra of a strongly sofic monoid with coefficients in an arbitrary field is always stably finite. Our results are extensions to strongly sofic monoids of two previously known properties of sofic groups. The first one is the celebrated Gromov-Weiss theorem asserting that every sofic group is surjunctive. The second is the Elek-Szab\'o theorem which says that group algebras of sofic groups satisfy Kaplansky's stable finiteness conjecture.

We investigate a family-nonuniversal Abelian extension of hypercharge, which significantly alters the phenomenological features of the standard model. Anomaly cancellation requires that the third quark family transforms differently from the first two quark families. Additionally, it acquires that three right-handed neutrinos are presented. This model generates naturally small neutrino masses and a $W$-boson mass deviation appropriate to recent measurements. Additionally, the model introduces flavor-changing neutral currents (FCNCs) of quarks coupled to the new gauge boson $Z'$ and new Higgs fields. These FCNCs significantly modify the neutral-meson mixing amplitudes and rare meson decays, which are studied in detail. We also address flavor changing processes in the charged lepton sector., Comment: 37 pages, 27 figures, 8 tables

Diffusion models excel at generating visually striking content from text but can inadvertently produce undesirable or harmful content when trained on unfiltered internet data. A practical solution is to selectively removing target concepts from the model, but this may impact the remaining concepts. Prior approaches have tried to balance this by introducing a loss term to preserve neutral content or a regularization term to minimize changes in the model parameters, yet resolving this trade-off remains challenging. In this work, we propose to identify and preserving concepts most affected by parameter changes, termed as \textit{adversarial concepts}. This approach ensures stable erasure with minimal impact on the other concepts. We demonstrate the effectiveness of our method using the Stable Diffusion model, showing that it outperforms state-of-the-art erasure methods in eliminating unwanted content while maintaining the integrity of other unrelated elements. Our code is available at \url{https://github.com/tuananhbui89/Erasing-Adversarial-Preservation}.

Service robots that work alongside humans in a shared environment need a navigation system that takes into account not only physical safety but also social norms for mutual cooperation. In this paper, we introduce a motion planning system that includes human states such as positions and velocities and their personal space for social-aware navigation. The system first extracts human positions from the LiDAR and the RGB-D camera. It then uses the Kalman filter to fuse that information for human state estimation. An asymmetric Gaussian function is then employed to model human personal space based on their states. This model is used as the input to the dynamic window approach algorithm to generate trajectories for the robot. Experiments show that the robot is able to navigate alongside humans in a dynamic environment while respecting their physical and psychological comfort., Comment: In Proceedings of 2024, the 7th International Conference on Control, Robotics and Informatics (ICCRI 2024)

Motion planning is an essential process for the navigation of unmanned aerial vehicles (UAVs) where they need to adapt to obstacles and different structures of their operating environment to reach the goal. This paper presents an optimal motion planner for UAVs operating in unknown complex environments. The motion planner receives point cloud data from a local range sensor and then converts it into a voxel grid representing the surrounding environment. A local trajectory guiding the UAV to the goal is then generated based on the voxel grid. This trajectory is further optimized using model predictive control (MPC) to enhance the safety, speed, and smoothness of UAV operation. The optimization is carried out via the definition of several cost functions and constraints, taking into account the UAV's dynamics and requirements. A number of simulations and comparisons with a state-of-the-art method have been conducted in a complex environment with many obstacles to evaluate the performance of our method. The results show that our method provides not only shorter and smoother trajectories but also faster and more stable speed profiles. It is also energy efficient making it suitable for various UAV applications., Comment: In proceedings of 2024, the 7th International Conference on Control, Robotics and Informatics (ICCRI 2024)

Drawing inspiration from human learning behaviors, this work proposes a novel approach to mitigate catastrophic forgetting in Prompt-based Continual Learning models by exploiting the relationships between continuously emerging class data. We find that applying human habits of organizing and connecting information can serve as an efficient strategy when training deep learning models. Specifically, by building a hierarchical tree structure based on the expanding set of labels, we gain fresh insights into the data, identifying groups of similar classes could easily cause confusion. Additionally, we delve deeper into the hidden connections between classes by exploring the original pretrained model's behavior through an optimal transport-based approach. From these insights, we propose a novel regularization loss function that encourages models to focus more on challenging knowledge areas, thereby enhancing overall performance. Experimentally, our method demonstrated significant superiority over the most robust state-of-the-art models on various benchmarks.

We introduce Flat Hilbert Bayesian Inference (FHBI), an algorithm designed to enhance generalization in Bayesian inference. Our approach involves an iterative two-step procedure with an adversarial functional perturbation step and a functional descent step within the reproducing kernel Hilbert spaces. This methodology is supported by a theoretical analysis that extends previous findings on generalization ability from finite-dimensional Euclidean spaces to infinite-dimensional functional spaces. To evaluate the effectiveness of FHBI, we conduct comprehensive comparisons against seven baseline methods on the VTAB-1K benchmark, which encompasses 19 diverse datasets across various domains with diverse semantics. Empirical results demonstrate that FHBI consistently outperforms the baselines by notable margins, highlighting its practical efficacy.

Reinforcement Learning from Human Feedback (RLHF) has greatly improved the performance of modern Large Language Models (LLMs). The RLHF process is resource-intensive and technically challenging, generally requiring a large collection of human preference labels over model-generated outputs. Reinforcement Learning from AI Feedback (RLAIF) addresses this data collection challenge by leveraging synthetic preferences generated by an LLM. However, recent work has shown that synthetic preferences labels may not align well with human preference judgments. To address this, we propose a hybrid approach that unifies RLHF and RLAIF methodologies. We introduce GenRM, an iterative algorithm that trains an LLM on self-generated reasoning traces, leading to synthetic preference labels matching human preference judgments. Empirically, we show that zero-shot LLM-based judgments under-perform compared to Bradley-Terry reward models on in-distribution tasks (between 9-36%). In contrast, GenRM achieves in-distribution accuracy comparable to Bradley-Terry models, while significantly outperforming them on out-of-distribution tasks (between 10-45%). Moreover, GenRM surpasses the performance of using LLMs as judges on both in-distribution (by 9-31%) and out-of-distribution tasks (by 2- 6%). Our results show that combining the strengths of RLHF and RLAIF offers a promising approach for improving the quality of synthetic preference labels.

Large Language Models (LLMs) have spurred interest in automatic evaluation methods for summarization, offering a faster, more cost-effective alternative to human evaluation. However, existing methods often fall short when applied to complex tasks like long-context summarizations and dialogue-based meeting summarizations. In this paper, we introduce CREAM (Comparison-Based Reference-Free Elo-Ranked Automatic Evaluation for Meeting Summarization), a novel framework that addresses the unique challenges of evaluating meeting summaries. CREAM leverages a combination of chain-of-thought reasoning and key facts alignment to assess conciseness and completeness of model-generated summaries without requiring reference. By employing an ELO ranking system, our approach provides a robust mechanism for comparing the quality of different models or prompt configurations.

The rapid expansion of online content has intensified the issue of information redundancy, underscoring the need for solutions that can identify genuinely new information. Despite this challenge, the research community has seen a decline in focus on novelty detection, particularly with the rise of large language models (LLMs). Additionally, previous approaches have relied heavily on human annotation, which is time-consuming, costly, and particularly challenging when annotators must compare a target document against a vast number of historical documents. In this work, we introduce NovAScore (Novelty Evaluation in Atomicity Score), an automated metric for evaluating document-level novelty. NovAScore aggregates the novelty and salience scores of atomic information, providing high interpretability and a detailed analysis of a document's novelty. With its dynamic weight adjustment scheme, NovAScore offers enhanced flexibility and an additional dimension to assess both the novelty level and the importance of information within a document. Our experiments show that NovAScore strongly correlates with human judgments of novelty, achieving a 0.626 Point-Biserial correlation on the TAP-DLND 1.0 dataset and a 0.920 Pearson correlation on an internal human-annotated dataset.

Despite its significant benefits in enhancing the transparency and trustworthiness of artificial intelligence (AI) systems, explainable AI (XAI) has yet to reach its full potential in real-world applications. One key challenge is that XAI can unintentionally provide adversaries with insights into black-box models, inevitably increasing their vulnerability to various attacks. In this paper, we develop a novel explanation-driven adversarial attack against black-box classifiers based on feature substitution, called XSub. The key idea of XSub is to strategically replace important features (identified via XAI) in the original sample with corresponding important features from a "golden sample" of a different label, thereby increasing the likelihood of the model misclassifying the perturbed sample. The degree of feature substitution is adjustable, allowing us to control how much of the original samples information is replaced. This flexibility effectively balances a trade-off between the attacks effectiveness and its stealthiness. XSub is also highly cost-effective in that the number of required queries to the prediction model and the explanation model in conducting the attack is in O(1). In addition, XSub can be easily extended to launch backdoor attacks in case the attacker has access to the models training data. Our evaluation demonstrates that XSub is not only effective and stealthy but also cost-effective, enabling its application across a wide range of AI models.

Assuming fundamental fermions possess a new Abelian gauge charge that depends on flavors of both quark and lepton, we obtain a simple extension of the Standard Model, which reveals some new physics insights. The new gauge charge anomaly cancellation not only explains the existence of just three fermion generations as observed but also requires the presence of a unique right-handed neutrino $\nu_R$ with a non-zero new gauge charge. Further, the new gauge charge breaking supplies a residual matter parity, under which the fundamental fermions and $\nu_R$ are even, whereas a right-handed neutrino $N_R$ without the new charge is odd. Consequently, light neutrino masses in our model are generated from the tree-level type-I seesaw mechanism induced by $\nu_R$ and from the one-loop scotogenic contribution accommodated by potential dark matter candidates, $N_R$ and dark scalars, odd under the matter parity. We examine new physics phenomena related to the additional gauge boson, which could be observed at colliders. We analyze the constraints imposed on our model by current experimental limits on neutrino masses, neutral meson oscillations, $B$-meson decays, and charged lepton flavor violating processes. We also investigate the potential dark matter candidates by considering relic density and direct detection., Comment: 38 pages, 10 figures, 5 tables

Diffusion models (DM) have become fundamental components of generative models, excelling across various domains such as image creation, audio generation, and complex data interpolation. Signal-to-Noise diffusion models constitute a diverse family covering most state-of-the-art diffusion models. While there have been several attempts to study Signal-to-Noise (S2N) diffusion models from various perspectives, there remains a need for a comprehensive study connecting different viewpoints and exploring new perspectives. In this study, we offer a comprehensive perspective on noise schedulers, examining their role through the lens of the signal-to-noise ratio (SNR) and its connections to information theory. Building upon this framework, we have developed a generalized backward equation to enhance the performance of the inference process.

Post-Training Quantization (PTQ) has received significant attention because it requires only a small set of calibration data to quantize a full-precision model, which is more practical in real-world applications in which full access to a large training set is not available. However, it often leads to overfitting on the small calibration dataset. Several methods have been proposed to address this issue, yet they still rely on only the calibration set for the quantization and they do not validate the quantized model due to the lack of a validation set. In this work, we propose a novel meta-learning based approach to enhance the performance of post-training quantization. Specifically, to mitigate the overfitting problem, instead of only training the quantized model using the original calibration set without any validation during the learning process as in previous PTQ works, in our approach, we both train and validate the quantized model using two different sets of images. In particular, we propose a meta-learning based approach to jointly optimize a transformation network and a quantized model through bi-level optimization. The transformation network modifies the original calibration data and the modified data will be used as the training set to learn the quantized model with the objective that the quantized model achieves a good performance on the original calibration data. Extensive experiments on the widely used ImageNet dataset with different neural network architectures demonstrate that our approach outperforms the state-of-the-art PTQ methods., Comment: Accepted by ECCV 2024

In [5], we have designed impulsive and feedback controls for harmonic chains with a point thermostat. In this work, we study the internal control for stochastic lattice dynamics, with the goal of controlling the transition kernel of the kinetic equation in the limit. A major novelty of the work is the introduction of a new geometric combinatorial argument, used to establish paths for the controls., Comment: 39 pages, 1 figure

We argue that the long-standing issues of neutrino mass and dark matter can be manifestly solved in a dark gauge symmetry $U(1)_D$ that transforms nontrivially only for three right-handed neutrinos $\nu_{1,2,3R}$ -- the counterparts of known left-handed neutrinos. This theory assigns $\nu_{1,2,3R}$ dark charge to be $D=0$, $-1$, and $+1$, respectively, in order for anomaly cancelation. Additionally, it imposes an inert Higgs doublet $\eta$ and two Higgs singlets $\xi,\phi$ with dark charge $D=+1$, $-1$, and $+2$, respectively. That said, the dark symmetry is broken by $\phi$ (by two units) down to a dark parity $P_D=(-1)^D$, for which $\nu_{2,3R}$ and $\eta,\xi$ are odd, whereas all other fields are even due to $D=0$. The lightest of these odd fields is stabilized by $P_D$, responsible for dark matter. Neutrino masses are generated by a scotoseesaw scheme, in which the seesaw part is mediated by $\nu_{1R}$, while the scotogenic part is mediated by $\nu_{2,3R}$, for which the hierarchy of atmospheric and solar neutrino mass splittings is explained., Comment: 14 pages, 2 figures, 1 table

Bayesian Neural Networks (BNNs) offer probability distributions for model parameters, enabling uncertainty quantification in predictions. However, they often underperform compared to deterministic neural networks. Utilizing mutual learning can effectively enhance the performance of peer BNNs. In this paper, we propose a novel approach to improve BNNs performance through deep mutual learning. The proposed approaches aim to increase diversity in both network parameter distributions and feature distributions, promoting peer networks to acquire distinct features that capture different characteristics of the input, which enhances the effectiveness of mutual learning. Experimental results demonstrate significant improvements in the classification accuracy, negative log-likelihood, and expected calibration error when compared to traditional mutual learning for BNNs., Comment: Accepted to NeurIPS 2023

Unpaired medical image synthesis aims to provide complementary information for an accurate clinical diagnostics, and address challenges in obtaining aligned multi-modal medical scans. Transformer-based models excel in imaging translation tasks thanks to their ability to capture long-range dependencies. Although effective in supervised training settings, their performance falters in unpaired image synthesis, particularly in synthesizing structural details. This paper empirically demonstrates that, lacking strong inductive biases, Transformer can converge to non-optimal solutions in the absence of paired data. To address this, we introduce UNet Structured Transformer (UNest), a novel architecture incorporating structural inductive biases for unpaired medical image synthesis. We leverage the foundational Segment-Anything Model to precisely extract the foreground structure and perform structural attention within the main anatomy. This guides the model to learn key anatomical regions, thus improving structural synthesis under the lack of supervision in unpaired training. Evaluated on two public datasets, spanning three modalities, i.e., MR, CT, and PET, UNest improves recent methods by up to 19.30% across six medical image synthesis tasks. Our code is released at https://github.com/HieuPhan33/MICCAI2024-UNest., Comment: MICCAI version before camera ready

Enhancing the electronic and diffusion properties of lithium-ion batteries is crucial for improving the performance of the fast-growing energy storage devices. Recently, fast-charging capability of commercial-like lithium-ion anodes with the least modification of the current manufactoring technology is of great interest. Here we use first principles methods with density functional theory and the climbing image-nudged elastic band method to evaluate the impact of an external electric field on the stability, electronic and diffusion properties of penta-graphene nanoribbons upon lithium adsorption. We show that by adsorbing a lithium atom, these semiconductor nanoribbons become metal with a formation energy of - 0.22 (eV). The lithium-ion mobility of this material is comparable to that of a common carbon graphite layer. Under a relatively small vertical electric field, the structural stability of these lithium-ion systems is even more stable, and their diffusion coefficient is enhanced significantly of ~719 times higher than that of the material in the absence of an applied electric field and ~521 times higher than in the case of commercial graphitic carbon layers. Our results highlight the role of an external electric field as a novel switch to improve the efficiency of lithium-ion batteries with penta-graphene nanoribbon electrodes and open a new horizon for the use of more environmentally friendly pentagonal materials as anode materials in lithium-ion battery industry., Comment: 21 pages, 5 figures

Topic modeling has been a widely used tool for unsupervised text analysis. However, comprehensive evaluations of a topic model remain challenging. Existing evaluation methods are either less comparable across different models (e.g., perplexity) or focus on only one specific aspect of a model (e.g., topic quality or document representation quality) at a time, which is insufficient to reflect the overall model performance. In this paper, we propose WALM (Words Agreement with Language Model), a new evaluation method for topic modeling that comprehensively considers the semantic quality of document representations and topics in a joint manner, leveraging the power of large language models (LLMs). With extensive experiments involving different types of topic models, WALM is shown to align with human judgment and can serve as a complementary evaluation method to the existing ones, bringing a new perspective to topic modeling. Our software package will be available at https://github.com/Xiaohao-Yang/Topic_Model_Evaluation, which can be integrated with many widely used topic models.

Scotogenic is a scheme for neutrino mass generation through the one-loop contribution of an inert scalar doublet and three sterile neutrinos. This work argues that such inert scalar doublet is a Goldstone boson mode associated with a gauge symmetry breaking. Hence, the resultant scotogenic gauge mechanism is very predictive, generating neutrino mass as contributed by a new gauge boson doublet that eats such Goldstone bosons. The dark matter stability is manifestly ensured by a matter parity as residual gauge symmetry for which a vector dark matter candidate is hinted., Comment: 6 pages, 3 figures

The multihop relay wireless networks have gained traction due to the emergence of Reconfigurable Intelligent Surfaces (RISs) which can be used as relays in high frequency range wireless network, including THz or mmWave. To select paths in these networks, the transmission performance plays the key network in these networks. In this paper, we enhance and greatly simplify the path selection in multihop relay RIS enabled wireless networks with what we refer to as interference graphs. Interference graphs are created based on SNR model, conical and cylindrical beam shapes in the transmission and the related interference model. Once created, they can be simply and efficiently used to select valid paths, without overestimation of the effect of interference. The results show that decreased ordering of conflict selections in the graphs yields the best results, as compared to conservative approach that tolerates no interference., Comment: This paper is uploaded here for research community, thus it is for non-commercial purposes

Sharpness-aware minimization (SAM) has been instrumental in improving deep neural network training by minimizing both the training loss and the sharpness of the loss landscape, leading the model into flatter minima that are associated with better generalization properties. In another aspect, Model-Agnostic Meta-Learning (MAML) is a framework designed to improve the adaptability of models. MAML optimizes a set of meta-models that are specifically tailored for quick adaptation to multiple tasks with minimal fine-tuning steps and can generalize well with limited data. In this work, we explore the connection between SAM and MAML in enhancing model generalization. We introduce Agnostic-SAM, a novel approach that combines the principles of both SAM and MAML. Agnostic-SAM adapts the core idea of SAM by optimizing the model toward wider local minima using training data, while concurrently maintaining low loss values on validation data. By doing so, it seeks flatter minima that are not only robust to small perturbations but also less vulnerable to data distributional shift problems. Our experimental results demonstrate that Agnostic-SAM significantly improves generalization over baselines across a range of datasets and under challenging conditions such as noisy labels or data limitation., Comment: Under review

Personalizing a large-scale pretrained Text-to-Image (T2I) diffusion model is challenging as it typically struggles to make an appropriate trade-off between its training data distribution and the target distribution, i.e., learning a novel concept with only a few target images to achieve personalization (aligning with the personalized target) while preserving text editability (aligning with diverse text prompts). In this paper, we propose PaRa, an effective and efficient Parameter Rank Reduction approach for T2I model personalization by explicitly controlling the rank of the diffusion model parameters to restrict its initial diverse generation space into a small and well-balanced target space. Our design is motivated by the fact that taming a T2I model toward a novel concept such as a specific art style implies a small generation space. To this end, by reducing the rank of model parameters during finetuning, we can effectively constrain the space of the denoising sampling trajectories towards the target. With comprehensive experiments, we show that PaRa achieves great advantages over existing finetuning approaches on single/multi-subject generation as well as single-image editing. Notably, compared to the prevailing fine-tuning technique LoRA, PaRa achieves better parameter efficiency (2x fewer learnable parameters) and much better target image alignment.

Despite the advances in text-to-image synthesis, particularly with diffusion models, generating visual instructions that require consistent representation and smooth state transitions of objects across sequential steps remains a formidable challenge. This paper introduces a simple, training-free framework to tackle the issues, capitalizing on the advancements in diffusion models and large language models (LLMs). Our approach systematically integrates text comprehension and image generation to ensure visual instructions are visually appealing and maintain consistency and accuracy throughout the instruction sequence. We validate the effectiveness by testing multi-step instructions and comparing the text alignment and consistency with several baselines. Our experiments show that our approach can visualize coherent and visually pleasing instructions, Comment: https://instruct-vis-zero.github.io/

A common type of zero-inflated data has certain true values incorrectly replaced by zeros due to data recording conventions (rare outcomes assumed to be absent) or details of data recording equipment (e.g. artificial zeros in gene expression data). Existing methods for zero-inflated data either fit the observed data likelihood via parametric mixture models that explicitly represent excess zeros, or aim to replace excess zeros by imputed values. If the goal of the analysis relies on knowing true data realizations, a particular challenge with zero-inflated data is identifiability, since it is difficult to correctly determine which observed zeros are real and which are inflated. This paper views zero-inflated data as a general type of missing data problem, where the observability indicator for a potentially censored variable is itself unobserved whenever a zero is recorded. We show that, without additional assumptions, target parameters involving a zero-inflated variable are not identified. However, if a proxy of the missingness indicator is observed, a modification of the effect restoration approach of Kuroki and Pearl allows identification and estimation, given the proxy-indicator relationship is known. If this relationship is unknown, our approach yields a partial identification strategy for sensitivity analysis. Specifically, we show that only certain proxy-indicator relationships are compatible with the observed data distribution. We give an analytic bound for this relationship in cases with a categorical outcome, which is sharp in certain models. For more complex cases, sharp numerical bounds may be computed using methods in Duarte et al.[2023]. We illustrate our method via simulation studies and a data application on central line-associated bloodstream infections (CLABSIs)., Comment: 28 pages, 8 figues, accepted for the 40th Conference on Uncertainty in Artificial Intelligence (UAI 2024)

A monoid $M$ is said to be surjunctive if every injective cellular automaton with finite alphabet over $M$ is surjective. We show that monoid algebras of surjunctive monoids are stably finite. In other words, given any field $K$ and any surjunctive monoid $M$, every one-sided invertible square matrix with entries in the monoid algebra $K[M]$ is two-sided invertible. Our proof uses first-order model theory., Comment: 18 page

The trend to equilibrium for reaction-diffusion systems coming out of chemistry is investigated, in the case when reaction processes might happen only on some open subsets of the domain. A special case has been studied recently in [Desvillettes, L., \\& Phung, K. D. (2022). Journal of Differential Equations, 338, 227-255] using log convexity technique from controllability theory, which in turn requires some amount of regularity for the solutions, and is difficult to generalise to more general systems. In this paper, we prove the convergence to equilibrium directly using vector-valued functional inequalities. One major advantage of our approach is that it allows to deal with nonlinearities of arbitrary orders, for which only global renormalised solutions are known to globally exist. For a specific situation where solutions are known to be bounded, we also prove the convergence to equilibrium when the diffusion as well as the reaction rates are degenerate. For this situation, we also treat the case of reactions happening in a set of strictly positive measure which may have an empty interior.

The Learning-to-match (LTM) framework proves to be an effective inverse optimal transport approach for learning the underlying ground metric between two sources of data, facilitating subsequent matching. However, the conventional LTM framework faces scalability challenges, necessitating the use of the entire dataset each time the parameters of the ground metric are updated. In adapting LTM to the deep learning context, we introduce the mini-batch Learning-to-match (m-LTM) framework for audio-text retrieval problems. This framework leverages mini-batch subsampling and Mahalanobis-enhanced family of ground metrics. Moreover, to cope with misaligned training data in practice, we propose a variant using partial optimal transport to mitigate the harm of misaligned data pairs in training data. We conduct extensive experiments on audio-text matching problems using three datasets: AudioCaps, Clotho, and ESC-50. Results demonstrate that our proposed method is capable of learning rich and expressive joint embedding space, which achieves SOTA performance. Beyond this, the proposed m-LTM framework is able to close the modality gap across audio and text embedding, which surpasses both triplet and contrastive loss in the zero-shot sound event detection task on the ESC-50 dataset. Notably, our strategy of employing partial optimal transport with m-LTM demonstrates greater noise tolerance than contrastive loss, especially under varying noise ratios in training data on the AudioCaps dataset. Our code is available at https://github.com/v-manhlt3/m-LTM-Audio-Text-Retrieval

This paper addresses the problem of controlling multiple unmanned aerial vehicles (UAVs) cooperating in a formation to carry out a complex task such as surface inspection. We first use the virtual leader-follower model to determine the topology and trajectory of the formation. A double-loop control system combining backstepping and sliding mode control techniques is then designed for the UAVs to track the trajectory. A radial basis function neural network (RBFNN) capable of estimating external disturbances is developed to enhance the robustness of the controller. The stability of the controller is proven by using the Lyapunov theorem. A number of comparisons and software-in-the-loop (SIL) tests have been conducted to evaluate the performance of the proposed controller. The results show that our controller not only outperforms other state-of-the-art controllers but is also sufficient for complex tasks of UAVs such as collecting surface data for inspection. The source code of our controller can be found at https://github.com/duynamrcv/rbf_bsmc

Generative models, e.g., Stable Diffusion, have enabled the creation of photorealistic images from text prompts. Yet, the generation of 360-degree panorama images from text remains a challenge, particularly due to the dearth of paired text-panorama data and the domain gap between panorama and perspective images. In this paper, we introduce a novel dual-branch diffusion model named PanFusion to generate a 360-degree image from a text prompt. We leverage the stable diffusion model as one branch to provide prior knowledge in natural image generation and register it to another panorama branch for holistic image generation. We propose a unique cross-attention mechanism with projection awareness to minimize distortion during the collaborative denoising process. Our experiments validate that PanFusion surpasses existing methods and, thanks to its dual-branch structure, can integrate additional constraints like room layout for customized panorama outputs. Code is available at https://chengzhag.github.io/publication/panfusion., Comment: CVPR 2024. Project Page: https://chengzhag.github.io/publication/panfusion Code: https://github.com/chengzhag/PanFusion

This paper extends Hopf-Galois theory to infinite field extensions and provides a natural definition of subextensions. For separable (possibly infinite) Hopf-Galois extensions, it provides a Galois correspondence. This correspondence also is a refinement of what was known in the case of finite separable Hopf-Galois extensions., Comment: 37 pages

We introduce Stable Code, the first in our new-generation of code language models series, which serves as a general-purpose base code language model targeting code completion, reasoning, math, and other software engineering-based tasks. Additionally, we introduce an instruction variant named Stable Code Instruct that allows conversing with the model in a natural chat interface for performing question-answering and instruction-based tasks. In this technical report, we detail the data and training procedure leading to both models. Their weights are available via Hugging Face for anyone to download and use at https://huggingface.co/stabilityai/stable-code-3b and https://huggingface.co/stabilityai/stable-code-instruct-3b. This report contains thorough evaluations of the models, including multilingual programming benchmarks, and the MT benchmark focusing on multi-turn dialogues. At the time of its release, Stable Code is the state-of-the-art open model under 3B parameters and even performs comparably to larger models of sizes 7 billion and 15 billion parameters on the popular Multi-PL benchmark. Stable Code Instruct also exhibits state-of-the-art performance on the MT-Bench coding tasks and on Multi-PL completion compared to other instruction tuned models. Given its appealing small size, we also provide throughput measurements on a number of edge devices. In addition, we open source several quantized checkpoints and provide their performance metrics compared to the original model.

Federated Class-Incremental Learning (FCIL) is an underexplored yet pivotal issue, involving the dynamic addition of new classes in the context of federated learning. In this field, Data-Free Knowledge Transfer (DFKT) plays a crucial role in addressing catastrophic forgetting and data privacy problems. However, prior approaches lack the crucial synergy between DFKT and the model training phases, causing DFKT to encounter difficulties in generating high-quality data from a non-anchored latent space of the old task model. In this paper, we introduce LANDER (Label Text Centered Data-Free Knowledge Transfer) to address this issue by utilizing label text embeddings (LTE) produced by pretrained language models. Specifically, during the model training phase, our approach treats LTE as anchor points and constrains the feature embeddings of corresponding training samples around them, enriching the surrounding area with more meaningful information. In the DFKT phase, by using these LTE anchors, LANDER can synthesize more meaningful samples, thereby effectively addressing the forgetting problem. Additionally, instead of tightly constraining embeddings toward the anchor, the Bounding Loss is introduced to encourage sample embeddings to remain flexible within a defined radius. This approach preserves the natural differences in sample embeddings and mitigates the embedding overlap caused by heterogeneous federated settings. Extensive experiments conducted on CIFAR100, Tiny-ImageNet, and ImageNet demonstrate that LANDER significantly outperforms previous methods and achieves state-of-the-art performance in FCIL. The code is available at https://github.com/tmtuan1307/lander., Comment: Accepted at CVPR 2024

We investigate a dark photon that arises from the UV model based upon $SU(3)_C\otimes SU(3)_L\otimes U(1)_X \otimes U(1)_G$ (3-3-1-1) gauge symmetry, where the last three factors enlarge the electroweak symmetry encompassing electric charge $Q=T_3 - 1/ \sqrt{3}T_8 +X$ and dark charge $D = -2/\sqrt{3} T_8 +G$. It is well-established that this model addresses the questions of family number, neutrino mass, and dark matter. It is shown in this work that if the 3-3-1-1 breaking scale is much bigger than the dark charge breaking scale, the relevant dark gauge boson $Z'$ is uniquely imprinted at TeV, avoiding dangerous FCNC processes, obeying precision electroweak measurements, as well as contributing to collider phenomena, even if no kinetic mixing is presented. The dark matter observables are perhaps governed by the dark charge breaking Higgs field instead of the dark photon., Comment: 16 pages, 3 figures, 2 tables; A scalar triplet relabelled for clarity

There has long been plenty of theoretical and empirical evidence supporting the success of ensemble learning. Deep ensembles in particular take advantage of training randomness and expressivity of individual neural networks to gain prediction diversity, ultimately leading to better generalization, robustness and uncertainty estimation. In respect of generalization, it is found that pursuing wider local minima result in models being more robust to shifts between training and testing sets. A natural research question arises out of these two approaches as to whether a boost in generalization ability can be achieved if ensemble learning and loss sharpness minimization are integrated. Our work investigates this connection and proposes DASH - a learning algorithm that promotes diversity and flatness within deep ensembles. More concretely, DASH encourages base learners to move divergently towards low-loss regions of minimal sharpness. We provide a theoretical backbone for our method along with extensive empirical evidence demonstrating an improvement in ensemble generalizability.

Diffusion models have shown remarkable capability in generating visually impressive content from textual descriptions. However, these models are trained on vast internet data, much of which contains undesirable elements such as sensitive content, copyrighted material, and unethical or harmful concepts. Therefore, beyond generating high-quality content, it is crucial to ensure these models do not propagate these undesirable elements. To address this issue, we propose a novel method to remove undesirable concepts from text-to-image diffusion models by incorporating a learnable prompt into the cross-attention module. This learnable prompt acts as additional memory, capturing the knowledge of undesirable concepts and reducing their dependency on the model parameters and corresponding textual inputs. By transferring this knowledge to the prompt, erasing undesirable concepts becomes more stable and has minimal negative impact on other concepts. We demonstrate the effectiveness of our method on the Stable Diffusion model, showcasing its superiority over state-of-the-art erasure methods in removing undesirable content while preserving unrelated elements.

Magnetic random access memory that uses magnetic tunnel junction memory cells is a high performance, non-volatile memory technology that goes beyond traditional charge-based memories. Today its speed is limited by the high magnetization of the memory storage layer. Here we show that fast and highly reliable switching is possible using a very low magnetization ferrimagnetic Heusler alloy, Mn3Ge. Moreover, the tunneling magnetoresistance is the highest yet achieved for a ferrimagnetic material at ambient temperature. Furthermore, the devices were prepared on technologically relevant amorphous substrates using a novel combination of a nitride seed layer and a chemical templating layer. These results show a clear path to the lowering of switching currents using ferrimagnetic Heusler materials and, therefore, to the scaling of high performance magnetic random access memories beyond those nodes possible with ferromagnetic devices., Comment: main manuscript 14 pages, 4 main figures and supplementary information. Submitted to Nature naotechnology

We establish generalizations of the well-known surjunctivity theorem of Gromov and Weiss as well as the dual-surjunctivity theorem of Capobianco, Kari and Taati for cellular automata (CA) to local perturbations of CA over sofic group universes. We also extend the results to a class of non-uniform cellular automata (NUCA) consisting of global perturbations with uniformly bounded singularity of CA. As an application, we obtain the surjunctivity of algebraic NUCA with uniformly bounded singularity over sofic groups. Moreover, we prove the stable finiteness of twisted group rings over sofic groups to generalize known results on Kaplansky's stable finiteness conjecture for group rings.

Knowledge distillation is an attractive approach for learning compact deep neural networks, which learns a lightweight student model by distilling knowledge from a complex teacher model. Attention-based knowledge distillation is a specific form of intermediate feature-based knowledge distillation that uses attention mechanisms to encourage the student to better mimic the teacher. However, most of the previous attention-based distillation approaches perform attention in the spatial domain, which primarily affects local regions in the input image. This may not be sufficient when we need to capture the broader context or global information necessary for effective knowledge transfer. In frequency domain, since each frequency is determined from all pixels of the image in spatial domain, it can contain global information about the image. Inspired by the benefits of the frequency domain, we propose a novel module that functions as an attention mechanism in the frequency domain. The module consists of a learnable global filter that can adjust the frequencies of student's features under the guidance of the teacher's features, which encourages the student's features to have patterns similar to the teacher's features. We then propose an enhanced knowledge review-based distillation model by leveraging the proposed frequency attention module. The extensive experiments with various teacher and student architectures on image classification and object detection benchmark datasets show that the proposed approach outperforms other knowledge distillation methods., Comment: Appear to WACV 2024