Search

Robotics can help address the growing worker shortage challenge of the manufacturing industry. As such, machine tending is a task collaborative robots can tackle that can also highly boost productivity. Nevertheless, existing robotics systems deployed in that sector rely on a fixed single-arm setup, whereas mobile robots can provide more flexibility and scalability. In this work, we introduce a multi-agent multi-machine tending learning framework by mobile robots based on Multi-agent Reinforcement Learning (MARL) techniques with the design of a suitable observation and reward. Moreover, an attention-based encoding mechanism is developed and integrated into Multi-agent Proximal Policy Optimization (MAPPO) algorithm to boost its performance for machine tending scenarios. Our model (AB-MAPPO) outperformed MAPPO in this new challenging scenario in terms of task success, safety, and resources utilization. Furthermore, we provided an extensive ablation study to support our various design decisions., Comment: 7 pages, 4 figures

Semantic role labeling is a crucial task in natural language processing, enabling better comprehension of natural language. However, the lack of annotated data in multiple languages has posed a challenge for researchers. To address this, a deep learning algorithm based on model transfer has been proposed. The algorithm utilizes a dataset consisting of the English portion of CoNLL2009 and a corpus of semantic roles in Persian. To optimize the efficiency of training, only ten percent of the educational data from each language is used. The results of the proposed model demonstrate significant improvements compared to Niksirt et al.'s model. In monolingual mode, the proposed model achieved a 2.05 percent improvement on F1-score, while in cross-lingual mode, the improvement was even more substantial, reaching 6.23 percent. Worth noting is that the compared model only trained two of the four stages of semantic role labeling and employed golden data for the remaining two stages. This suggests that the actual superiority of the proposed model surpasses the reported numbers by a significant margin. The development of cross-lingual methods for semantic role labeling holds promise, particularly in addressing the scarcity of annotated data for various languages. These advancements pave the way for further research in understanding and processing natural language across different linguistic contexts.

Differentially private stochastic gradient descent (DP-SGD) has been instrumental in privately training deep learning models by providing a framework to control and track the privacy loss incurred during training. At the core of this computation lies a subsampling method that uses a privacy amplification lemma to enhance the privacy guarantees provided by the additive noise. Fixed size subsampling is appealing for its constant memory usage, unlike the variable sized minibatches in Poisson subsampling. It is also of interest in addressing class imbalance and federated learning. However, the current computable guarantees for fixed-size subsampling are not tight and do not consider both add/remove and replace-one adjacency relationships. We present a new and holistic R{\'e}nyi differential privacy (RDP) accountant for DP-SGD with fixed-size subsampling without replacement (FSwoR) and with replacement (FSwR). For FSwoR we consider both add/remove and replace-one adjacency. Our FSwoR results improves on the best current computable bound by a factor of $4$. We also show for the first time that the widely-used Poisson subsampling and FSwoR with replace-one adjacency have the same privacy to leading order in the sampling probability. Accordingly, our work suggests that FSwoR is often preferable to Poisson subsampling due to constant memory usage. Our FSwR accountant includes explicit non-asymptotic upper and lower bounds and, to the authors' knowledge, is the first such analysis of fixed-size RDP with replacement for DP-SGD. We analytically and empirically compare fixed size and Poisson subsampling, and show that DP-SGD gradients in a fixed-size subsampling regime exhibit lower variance in practice in addition to memory usage benefits., Comment: 39 pages, 10 figures

While deep face recognition models have demonstrated remarkable performance, they often struggle on the inputs from domains beyond their training data. Recent attempts aim to expand the training set by relying on computationally expensive and inherently challenging image-space augmentation of image generation modules. In an orthogonal direction, we present a simple yet effective method to expand the training data by interpolating between instance-level feature statistics across labeled and unlabeled sets. Our method, dubbed Targeted Style Adversary (TSA), is motivated by two observations: (i) the input domain is reflected in feature statistics, and (ii) face recognition model performance is influenced by style information. Shifting towards an unlabeled style implicitly synthesizes challenging training instances. We devise a recognizability metric to constraint our framework to preserve the inherent identity-related information of labeled instances. The efficacy of our method is demonstrated through evaluations on unconstrained benchmarks, outperforming or being on par with its competitors while offering nearly a 70\% improvement in training speed and 40\% less memory consumption.

Reinforcement Learning (RL) is a branch of Artificial Intelligence (AI) which focuses on training agents to make decisions by interacting with their environment to maximize cumulative rewards. An overview of RL is provided in this paper, which discusses its core concepts, methodologies, recent trends, and resources for learning. We provide a detailed explanation of key components of RL such as states, actions, policies, and reward signals so that the reader can build a foundational understanding. The paper also provides examples of various RL algorithms, including model-free and model-based methods. In addition, RL algorithms are introduced and resources for learning and implementing them are provided, such as books, courses, and online communities. This paper demystifies a comprehensive yet simple introduction for beginners by offering a structured and clear pathway for acquiring and implementing real-time techniques.

Multimodal Large Language Models (MLLMs) demonstrate remarkable image-language capabilities, but their widespread use faces challenges in cost-effective training and adaptation. Existing approaches often necessitate expensive language model retraining and limited adaptability. Additionally, the current focus on zero-shot performance improvements offers insufficient guidance for task-specific tuning. We propose CROME, an efficient vision-language instruction tuning framework. It features a novel gated cross-modal adapter that effectively combines visual and textual representations prior to input into a frozen LLM. This lightweight adapter, trained with minimal parameters, enables efficient cross-modal understanding. Notably, CROME demonstrates superior zero-shot performance on standard visual question answering and instruction-following benchmarks. Moreover, it yields fine-tuning with exceptional parameter efficiency, competing with task-specific specialist state-of-the-art methods. CROME demonstrates the potential of pre-LM alignment for building scalable, adaptable, and parameter-efficient multimodal models.

Despite their recent successes, Transformer-based large language models show surprising failure modes. A well-known example of such failure modes is their inability to length-generalize: solving problem instances at inference time that are longer than those seen during training. In this work, we further explore the root cause of this failure by performing a detailed analysis of model behaviors on the simple parity task. Our analysis suggests that length generalization failures are intricately related to a model's inability to perform random memory accesses within its context window. We present supporting evidence for this hypothesis by demonstrating the effectiveness of methodologies that circumvent the need for indexing or that enable random token access indirectly, through content-based addressing. We further show where and how the failure to perform random memory access manifests through attention map visualizations., Comment: Published as a conference paper at COLM 2024

Applying the modified Barrow entropy, inspired by the quantum fluctuation effects, to the cosmological background, and using thermodynamics-gravity conjuncture, the Friedmann equations get modified as well. In this paper, we explore the holographic dark energy with Granda-Oliveros (GO) IR cutoff, in the context of the modified Barrow cosmology. First, we assume two dark components of the universe evolves independently and obtain the cosmological parameters and explore the cosmic evolution. Second, we consider an interaction term between dark energy (DE) and dark matter (DM). We observe that the Barrow parameter $\delta$ crucially affects the cosmic dynamics, causes the transition from the decelerating phase to the accelerating phase occurs later. We find out that the equation of state parameter is in the quintessence region in the past and crosses the phantom divide at the present time. Finally, we examine the squared speed of sound analysis for this model. According to the squared sound speed diagrams, the results indicate that the presence of interaction between DM and DE as well as increasing in the value of $\delta$ leads to the manifestation of signs of instability in the past $(v_s^2<0)$. Furthermore, by examining the statefinder, we find that presence of $\delta$ also makes a distinction between holographic dark energy in Barrow cosmology with GO-IR cutoff and the $\Lambda$CDM model. In fact, increasing $\delta$ causes the statefinder diagram move away from the point of $\left\lbrace r,s\right\rbrace= \left\lbrace 1,0\right\rbrace$ at $z=0$., Comment: 20 pages, 11 figures

Offline reinforcement learning has shown promise for solving tasks in safety-critical settings, such as clinical decision support. Its application, however, has been limited by the lack of interpretability and interactivity for clinicians. To address these challenges, we propose the medical decision transformer (MeDT), a novel and versatile framework based on the goal-conditioned reinforcement learning paradigm for sepsis treatment recommendation. MeDT uses the decision transformer architecture to learn a policy for drug dosage recommendation. During offline training, MeDT utilizes collected treatment trajectories to predict administered treatments for each time step, incorporating known treatment outcomes, target acuity scores, past treatment decisions, and current and past medical states. This analysis enables MeDT to capture complex dependencies among a patient's medical history, treatment decisions, outcomes, and short-term effects on stability. Our proposed conditioning uses acuity scores to address sparse reward issues and to facilitate clinician-model interactions, enhancing decision-making. Following training, MeDT can generate tailored treatment recommendations by conditioning on the desired positive outcome (survival) and user-specified short-term stability improvements. We carry out rigorous experiments on data from the MIMIC-III dataset and use off-policy evaluation to demonstrate that MeDT recommends interventions that outperform or are competitive with existing offline reinforcement learning methods while enabling a more interpretable, personalized and clinician-directed approach.

Feature selection plays a pivotal role in the data preprocessing and model-building pipeline, significantly enhancing model performance, interpretability, and resource efficiency across diverse domains. In population-based optimization methods, the generation of diverse individuals holds utmost importance for adequately exploring the problem landscape, particularly in highly multi-modal multi-objective optimization problems. Our study reveals that, in line with findings from several prior research papers, commonly employed crossover and mutation operations lack the capability to generate high-quality diverse individuals and tend to become confined to limited areas around various local optima. This paper introduces an augmentation to the diversity of the population in the well-established multi-objective scheme of the genetic algorithm, NSGA-II. This enhancement is achieved through two key components: the genuine initialization method and the substitution of the worst individuals with new randomly generated individuals as a re-initialization approach in each generation. The proposed multi-objective feature selection method undergoes testing on twelve real-world classification problems, with the number of features ranging from 2,400 to nearly 50,000. The results demonstrate that replacing the last front of the population with an equivalent number of new random individuals generated using the genuine initialization method and featuring a limited number of features substantially improves the population's quality and, consequently, enhances the performance of the multi-objective algorithm., Comment: 8 pages, 3 figures, published in IEEE WCCI 2024 conference, DOI added

Higher-order notions of Kreweras complementation have appeared in the literature in the works of Krawczyk, Speicher, Mastnak, Nica, Arizmendi, Vargas, and others. While the theory has been developed primarily for specific applications in free probability, it also possesses an elegant, purely combinatorial core that is of independent interest. The present article aims at offering a simple account of various aspects of higher-order Kreweras complementation on the basis of elementary arithmetic, (co)algebraic, categorical and simplicial properties of noncrossing partitions. The main idea is to see noncrossing partitions as providing an interesting noncommutative analogue of the interplay between the divisibility poset and the multiplicative monoid of positive integers. Just as the divisibility poset can be regarded as the decalage of the multiplicative monoid, we exhibit the lattice of noncrossing partitions as the decalage of a partial monoid structure on noncrossing partitions encoding higher-order Kreweras complements. While our results may be considered known, some of the viewpoints can be regarded as novel, providing an efficient approach both conceptually and computationally., Comment: For the CATMI proceedings volume. 19pp

Aiming to enhance Face Recognition (FR) on Low-Quality (LQ) inputs, recent studies suggest incorporating synthetic LQ samples into training. Although promising, the quality factors that are considered in these works are general rather than FR-specific, \eg, atmospheric turbulence, resolution, \etc. Motivated by the observation of the vulnerability of current FR models to even small Face Alignment Errors (FAE) in LQ images, we present a simple yet effective method that considers FAE as another quality factor that is tailored to FR. We seek to improve LQ FR by enhancing FR models' robustness to FAE. To this aim, we formalize the problem as a combination of differentiable spatial transformations and adversarial data augmentation in FR. We perturb the alignment of the training samples using a controllable spatial transformation and enrich the training with samples expressing FAE. We demonstrate the benefits of the proposed method by conducting evaluations on IJB-B, IJB-C, IJB-S (+4.3\% Rank1), and TinyFace (+2.63\%). \href{https://github.com/msed-Ebrahimi/ARoFace}{https://github.com/msed-Ebrahimi/ARoFace}, Comment: European Conference on Computer Vision (ECCV 2024)

Semantic Textual Relatedness holds significant relevance in Natural Language Processing, finding applications across various domains. Traditionally, approaches to STR have relied on knowledge-based and statistical methods. However, with the emergence of Large Language Models, there has been a paradigm shift, ushering in new methodologies. In this paper, we delve into the investigation of sentence-level STR within Track A (Supervised) by leveraging fine-tuning techniques on the RoBERTa transformer. Our study focuses on assessing the efficacy of this approach across different languages. Notably, our findings indicate promising advancements in STR performance, particularly in Latin languages. Specifically, our results demonstrate notable improvements in English, achieving a correlation of 0.82 and securing a commendable 19th rank. Similarly, in Spanish, we achieved a correlation of 0.67, securing the 15th position. However, our approach encounters challenges in languages like Arabic, where we observed a correlation of only 0.38, resulting in a 20th rank., Comment: 10 pages, 9 figures, 4 tables

Detecting Machine-Generated Text (MGT) has emerged as a significant area of study within Natural Language Processing. While language models generate text, they often leave discernible traces, which can be scrutinized using either traditional feature-based methods or more advanced neural language models. In this research, we explore the effectiveness of fine-tuning a RoBERTa-base transformer, a powerful neural architecture, to address MGT detection as a binary classification task. Focusing specifically on Subtask A (Monolingual-English) within the SemEval-2024 competition framework, our proposed system achieves an accuracy of 78.9% on the test dataset, positioning us at 57th among participants. Our study addresses this challenge while considering the limited hardware resources, resulting in a system that excels at identifying human-written texts but encounters challenges in accurately discerning MGTs., Comment: 8 pages, 3 figures, 2 tables. Proceedings of the 18th International Workshop on Semantic Evaluation (SemEval-2024)

We argue that Content-based filtering (CBF) and Graph-based methods (GB) complement one another in Academic Search recommendations. The scientific literature can be viewed as a conversation between authors and the audience. CBF uses abstracts to infer authors' positions, and GB uses citations to infer responses from the audience. In this paper, we describe nine differences between CBF and GB, as well as synergistic opportunities for hybrid combinations. Two embeddings will be used to illustrate these opportunities: (1) Specter, a CBF method based on BERT-like deepnet encodings of abstracts, and (2) ProNE, a GB method based on spectral clustering of more than 200M papers and 2B citations from Semantic Scholar.

Over recent years, deep convolutional neural networks have significantly advanced the field of face recognition techniques for both verification and identification purposes. Despite the impressive accuracy, these neural networks are often criticized for lacking explainability. There is a growing demand for understanding the decision-making process of AI-based face recognition systems. Some studies have investigated the use of visual saliency maps as explanations, but they have predominantly focused on the specific face verification case. The discussion on more general face recognition scenarios and the corresponding evaluation methodology for these explanations have long been absent in current research. Therefore, this manuscript conceives a comprehensive explanation framework for face recognition tasks. Firstly, an exhaustive definition of visual saliency map-based explanations for AI-based face recognition systems is provided, taking into account the two most common recognition situations individually, i.e., face verification and identification. Secondly, a new model-agnostic explanation method named CorrRISE is proposed to produce saliency maps, which reveal both the similar and dissimilar regions between any given face images. Subsequently, the explanation framework conceives a new evaluation methodology that offers quantitative measurement and comparison of the performance of general visual saliency explanation methods in face recognition. Consequently, extensive experiments are carried out on multiple verification and identification scenarios. The results showcase that CorrRISE generates insightful saliency maps and demonstrates superior performance, particularly in similarity maps in comparison with the state-of-the-art explanation approaches., Comment: arXiv admin note: text overlap with arXiv:2305.08546

Protein sequence design, determined by amino acid sequences, are essential to protein engineering problems in drug discovery. Prior approaches have resorted to evolutionary strategies or Monte-Carlo methods for protein design, but often fail to exploit the structure of the combinatorial search space, to generalize to unseen sequences. In the context of discrete black box optimization over large search spaces, learning a mutation policy to generate novel sequences with reinforcement learning is appealing. Recent advances in protein language models (PLMs) trained on large corpora of protein sequences offer a potential solution to this problem by scoring proteins according to their biological plausibility (such as the TM-score). In this work, we propose to use PLMs as a reward function to generate new sequences. Yet the PLM can be computationally expensive to query due to its large size. To this end, we propose an alternative paradigm where optimization can be performed on scores from a smaller proxy model that is periodically finetuned, jointly while learning the mutation policy. We perform extensive experiments on various sequence lengths to benchmark RL-based approaches, and provide comprehensive evaluations along biological plausibility and diversity of the protein. Our experimental results include favorable evaluations of the proposed sequences, along with high diversity scores, demonstrating that RL is a strong candidate for biological sequence design. Finally, we provide a modular open source implementation can be easily integrated in most RL training loops, with support for replacing the reward model with other PLMs, to spur further research in this domain. The code for all experiments is provided in the supplementary material., Comment: 22 pages, 7 figures, 4 tables

To achieve state-of-the-art jet energy resolution for Particle Flow, sophisticated energy clustering algorithms must be developed that can fully exploit available information to separate energy deposits from charged and neutral particles. Three published neural network-based shower separation models were applied to simulation and experimental data to measure the performance of the highly granular CALICE Analogue Hadronic Calorimeter (AHCAL) technological prototype in distinguishing the energy deposited by a single charged and single neutral hadron for Particle Flow. The performance of models trained using only standard spatial and energy and charged track position information from an event was compared to models trained using timing information available from AHCAL, which is expected to improve sensitivity to shower development and, therefore, aid in clustering. Both simulation and experimental data were used to train and test the models and their performances were compared. The best-performing neural network achieved significantly superior event reconstruction when timing information was utilised in training for the case where the charged hadron had more energy than the neutral one, motivating temporally sensitive calorimeters. All models under test were observed to tend to allocate energy deposited by the more energetic of the two showers to the less energetic one. Similar shower reconstruction performance was observed for a model trained on simulation and applied to data and a model trained and applied to data.

In the last decade, the concept of path signature has found great success in data science applications, where it provides features describing the path. This is partly explained by the fact that it is possible to compute the signature of a path in linear time, owing to a dynamic programming principle, based on Chen's identity. The path signature can be regarded as a specific example of product or time-/path-ordered integral. In other words, it can be seen as a 1-parameter object build on iterated integrals over a path. Increasing the number of parameters by one, which amounts to considering iterated integrals over surfaces, is more complicated. An observation that is familiar in the context of higher gauge theory where multiparameter iterated integrals play an important role. The 2-parameter case is naturally related to a non-commutative version of Stokes' theorem, which is understood to be fundamentally linked to the concept of crossed modules of groups. Indeed, crossed modules with non-trivial kernel of the feedback map permit to compute features of a surface that go beyond what can be expressed by computing line integrals along the boundary of a surface. A good candidate for the crossed analog of free Lie algebra then seems to be a certain free crossed module over it. Building on work by Kapranov, we study the analog to the classical path signature taking values in such a free crossed module of Lie algebra. In particular, we provide a Magnus-type expression for the logarithm of surface signature as well as a sewing lemma for the crossed module setting.

The present study addresses the challenge of enhancing computational efficiency without compromising accuracy in numerical simulations of vacuum gas dynamics using the direct simulation Monte Carlo (DSMC) method. A technique termed "fixed particle per cell (FPPC)" was employed, which enforces a fixed number of simulator particles across all computational cells. The proposed technique eliminates the need for real-time adjustment of particle weights during simulation, reducing calculation time. Using the SPARTA solver, simulations of rarefied gas flow in a micromixer and rarefied supersonic airflow around a cylinder were conducted to validate the proposed technique. Results demonstrate that applying the FPPC technique effectively reduces computational costs while yielding results comparable to conventional DSMC implementations. Additionally, the application of local grid refinement coupled with the FPPC technique was investigated. The results show that integrating local grid refinement with the FPPC technique enables accurate prediction of flow behaviour in regions with significant gradients. These findings highlight the efficacy of the proposed technique in improving the accuracy and efficiency of numerical simulations of complex vacuum gas dynamics at a reduced computational cost.

A singular graph is a finite graph $\Gamma$ whose adjacency matrix is singular. Let $G$ be a finite group and $p$ a prime divisor of the order of $G$. Also let $\Omega(G,p)$ be the set of all $p$-singular elements of $G$. In this paper, we show that the Cayley graph $\Gamma_p(G):=\mathrm{Cay}(G,\Omega(G,p))$ is a singular graph, if and only if the group $G$ has a non-principal $p$-block. We apply this result to obtain new classes of singular Cayley graphs (See Corollary 1.3). We also prove that if $G$ is $p$-solvable, then the diameter of $\Gamma_p(G)$ is at most the $p$-part of the order of $G$.

The future 6G network is envisioned to be AI-native, and as such, ML models will be pervasive in support of optimizing performance, reducing energy consumption, and in coping with increasing complexity and heterogeneity. A key challenge is automating the process of finding optimal model architectures satisfying stringent requirements stemming from varying tasks, dynamicity and available resources in the infrastructure and deployment positions. In this paper, we describe and review the state-of-the-art in Neural Architecture Search and Transfer Learning and their applicability in networking. Further, we identify open research challenges and set directions with a specific focus on three main requirements with elements unique to the future network, namely combining NAS and TL, multi-objective search, and tabular data. Finally, we outline and discuss both near-term and long-term work ahead.

Model-based reinforcement learning agents utilizing transformers have shown improved sample efficiency due to their ability to model extended context, resulting in more accurate world models. However, for complex reasoning and planning tasks, these methods primarily rely on continuous representations. This complicates modeling of discrete properties of the real world such as disjoint object classes between which interpolation is not plausible. In this work, we introduce discrete abstract representations for transformer-based learning (DART), a sample-efficient method utilizing discrete representations for modeling both the world and learning behavior. We incorporate a transformer-decoder for auto-regressive world modeling and a transformer-encoder for learning behavior by attending to task-relevant cues in the discrete representation of the world model. For handling partial observability, we aggregate information from past time steps as memory tokens. DART outperforms previous state-of-the-art methods that do not use look-ahead search on the Atari 100k sample efficiency benchmark with a median human-normalized score of 0.790 and beats humans in 9 out of 26 games. We release our code at https://pranaval.github.io/DART/., Comment: Accepted at ICML 2024

This paper examines the alignment of inductive biases in machine learning (ML) with structural models of economic dynamics. Unlike dynamical systems found in physical and life sciences, economics models are often specified by differential equations with a mixture of easy-to-enforce initial conditions and hard-to-enforce infinite horizon boundary conditions (e.g. transversality and no-ponzi-scheme conditions). Traditional methods for enforcing these constraints are computationally expensive and unstable. We investigate algorithms where those infinite horizon constraints are ignored, simply training unregularized kernel machines and neural networks to obey the differential equations. Despite the inherent underspecification of this approach, our findings reveal that the inductive biases of these ML models innately enforce the infinite-horizon conditions necessary for the well-posedness. We theoretically demonstrate that (approximate or exact) min-norm ML solutions to interpolation problems are sufficient conditions for these infinite-horizon boundary conditions in a wide class of problems. We then provide empirical evidence that deep learning and ridgeless kernel methods are not only theoretically sound with respect to economic assumptions, but may even dominate classic algorithms in low to medium dimensions. More importantly, these results give confidence that, despite solving seemingly ill-posed problems, there are reasons to trust the plethora of black-box ML algorithms used by economists to solve previously intractable, high-dimensional dynamical systems -- paving the way for future work on estimation of inverse problems with embedded optimal control problems.

Commercially impactful quantum algorithms such as quantum chemistry and Shor's algorithm require a number of qubits and gates far beyond the capacity of any existing quantum processor. Distributed architectures, which scale horizontally by networking modules, provide a route to commercial utility and will eventually surpass the capability of any single quantum computing module. Such processors consume remote entanglement distributed between modules to realize distributed quantum logic. Networked quantum computers will therefore require the capability to rapidly distribute high fidelity entanglement between modules. Here we present preliminary demonstrations of some key distributed quantum computing protocols on silicon T centres in isotopically-enriched silicon. We demonstrate the distribution of entanglement between modules and consume it to apply a teleported gate sequence, establishing a proof-of-concept for T centres as a distributed quantum computing and networking platform., Comment: 14 pages, 13 figures

Few-shot learning aims to learn representations that can tackle novel tasks given a small number of examples. Recent studies show that cross-modal learning can improve representations for few-shot classification. More specifically, language is a rich modality that can be used to guide visual learning. In this work, we experiment with a multi-modal architecture for few-shot learning that consists of three components: a classifier, an auxiliary network, and a bridge network. While the classifier performs the main classification task, the auxiliary network learns to predict language representations from the same input, and the bridge network transforms high-level features of the auxiliary network into modulation parameters for layers of the few-shot classifier using conditional batch normalization. The bridge should encourage a form of lightweight semantic alignment between language and vision which could be useful for the classifier. However, after evaluating the proposed approach on two popular few-shot classification benchmarks we find that a) the improvements do not reproduce across benchmarks, and b) when they do, the improvements are due to the additional compute and parameters introduced by the bridge network. We contribute insights and recommendations for future work in multi-modal meta-learning, especially when using language representations.

Despite their remarkable progress, Multimodal Large Language Models (MLLMs) tend to hallucinate factually inaccurate information. In this work, we address object hallucinations in MLLMs, where information is offered about an object that is not present in the model input. We introduce a contrastive tuning method that can be applied to a pretrained off-the-shelf MLLM for mitigating hallucinations while preserving its general vision-language capabilities. For a given factual token, we create a hallucinated token through generative data augmentation by selectively altering the ground-truth information. The proposed contrastive tuning is applied at the token level to improve the relative likelihood of the factual token compared to the hallucinated one. Our thorough evaluation confirms the effectiveness of contrastive tuning in mitigating hallucination. Moreover, the proposed contrastive tuning is simple, fast, and requires minimal training with no additional overhead at inference.

According to the thermodynamics-gravity conjecture, any modification to the entropy expression leads to the modified cosmological field equations. Based on this, we investigate the cosmological consequences of the modified Friedmann equations when the entropy associated with the horizon is in the form of Tsallis/Barrow entropy and the dark energy (DE) component is in the form of ghost dark energy (GDE). We perform a dynamical system analysis and see that the Tsallis GDE(TGDE) and Barrow GDE(BGDE) can exhibit a correct phase space evolution for suitable range of the free parameters($0<\Delta<1$ for BGDE and $\beta<3/2$ for TGDE). We find that in BGDE and TGDE (with $Q=3b^2 H(\rho_D+\rho_m)$), there exist an early radiation dominated phase of expansion which is absent for GDE model in standard cosmology. It is worth mentioning that seeking an unstable phase of matter dominated in TGDE leads to a new constraint on $\beta$($<3/2$). Using the resulting range of free parameters from latter step we find that both models are capable to explain the cosmic evolution from deceleration to an accelerated phase. We observe that increasing $\Delta$ ($\beta$) parameters leads to a delay in the cosmic phase transition. We conclude that BGDE and TGDE are viable cosmological models which predict a consistent phase transition of the cosmic expansion for suitable ranges of the parameters. We also calculate the squared sound speed for both models and find out that they are unstable against perturbations..., Comment: 14 pages, 31 figures

The Active Matter Evaluation Package (AMEP) is a Python library for analyzing simulation data of particle-based and continuum simulations. It provides a powerful and simple interface for handling large data sets and for calculating and visualizing a broad variety of observables that are relevant to active matter systems. Examples range from the mean-square displacement and the structure factor to cluster-size distributions, binder cumulants, and growth exponents. AMEP is written in pure Python and is based on powerful libraries such as NumPy, SciPy, Matplotlib, and scikit-image. Computationally expensive methods are parallelized and optimized to run efficiently on workstations, laptops, and high-performance computing architectures, and an HDF5-based data format is used in the backend to store and handle simulation data as well as analysis results. AMEP provides the first comprehensive framework for analyzing simulation results of both particle-based and continuum simulations (as well as experimental data) of active matter systems. In particular, AMEP also allows it to analyze simulations that combine particle-based and continuum techniques such as used to study the motion of bacteria in chemical fields or for modeling particle motion in a flow field. AMEP is available at https://amepproject.de and can be installed via conda and pip., Comment: See https://github.com/amepproject/amep for the source code and https://amepproject.de/ for the documentation

The recent success of large language models (LLMs) trained on static, pre-collected, general datasets has sparked numerous research directions and applications. One such direction addresses the non-trivial challenge of integrating pre-trained LLMs into dynamic data distributions, task structures, and user preferences. Pre-trained LLMs, when tailored for specific needs, often experience significant performance degradation in previous knowledge domains -- a phenomenon known as "catastrophic forgetting". While extensively studied in the continual learning (CL) community, it presents new manifestations in the realm of LLMs. In this survey, we provide a comprehensive overview of the current research progress on LLMs within the context of CL. This survey is structured into four main sections: we first describe an overview of continually learning LLMs, consisting of two directions of continuity: vertical continuity (or vertical continual learning), i.e., continual adaptation from general to specific capabilities, and horizontal continuity (or horizontal continual learning), i.e., continual adaptation across time and domains (Section 3). We then summarize three stages of learning LLMs in the context of modern CL: Continual Pre-Training (CPT), Domain-Adaptive Pre-training (DAP), and Continual Fine-Tuning (CFT) (Section 4). Then we provide an overview of evaluation protocols for continual learning with LLMs, along with the current available data sources (Section 5). Finally, we discuss intriguing questions pertaining to continual learning for LLMs (Section 6). The full list of papers examined in this survey is available at https://github.com/Wang-ML-Lab/llm-continual-learning-survey., Comment: 47 pages, 2 figures, 4 tables. Work in progress

The extensive scope of large language models (LLMs) across various domains underscores the critical importance of responsibility in their application, beyond natural language processing. In particular, the randomized nature of LLMs, coupled with inherent biases and historical stereotypes in data, raises critical concerns regarding reliability and equity. Addressing these challenges are necessary before using LLMs for applications with societal impact. Towards addressing this gap, we introduce REQUAL-LM, a novel method for finding reliable and equitable LLM outputs through aggregation. Specifically, we develop a Monte Carlo method based on repeated sampling to find a reliable output close to the mean of the underlying distribution of possible outputs. We formally define the terms such as reliability and bias, and design an equity-aware aggregation to minimize harmful bias while finding a highly reliable output. REQUAL-LM does not require specialized hardware, does not impose a significant computing load, and uses LLMs as a blackbox. This design choice enables seamless scalability alongside the rapid advancement of LLM technologies. Our system does not require retraining the LLMs, which makes it deployment ready and easy to adapt. Our comprehensive experiments using various tasks and datasets demonstrate that REQUAL- LM effectively mitigates bias and selects a more equitable response, specifically the outputs that properly represents minority groups.

Engineering superlattices (SLs) - which are spatially periodic potential landscapes for electrons - is an emerging approach for the realization of exotic properties, including superconductivity and correlated insulators, in two-dimensional materials. While moir\'e SL engineering has been a popular approach, nanopatterning is an attractive alternative offering control over the pattern and wavelength of the SL. However, the disorder arising in the system due to imperfect nanopatterning is seldom studied. Here, by creating a square lattice of nanoholes in the $SiO_2$ dielectric layer using nanolithography, we study the superlattice potential and the disorder formed in hBN-graphene-hBN heterostructures. Specifically, we observe that while electrical transport shows distinct superlattice satellite peaks, the disorder of the device is significantly higher than graphene devices without any SL. We use finite-element simulations combined with a resistor network model to calculate the effects of this disorder on the transport properties of graphene. We consider three types of disorder: nanohole size variations, adjacent nanohole mergers, and nanohole vacancies. Comparing our experimental results with the model, we find that the disorder primarily originates from nanohole size variations rather than nanohole mergers in square SLs. We further confirm the validity of our model by comparing the results with quantum transport simulations. Our findings highlight the applicability of our simple framework to predict and engineer disorder in patterned SLs, specifically correlating variations in the resultant SL patterns to the observed disorder. Our combined experimental and theoretical results could serve as a valuable guide for optimizing nanofabrication processes to engineer disorder in nanopatterned SLs.

Chemical interface damping is a change in the effective collision frequency of conduction band electrons in metal originating from a chemical change of the metal interface. In this work, we present in-situ ellipsometric measurements that reveal the chemical interface damping effect from electrochemical oxidation of single crystal and polycrystalline gold films. We observe an increase in collision frequency of up to 21 meV for single-crystalline gold. To compare to results obtained with thiols and metal-oxides on gold nanoparticles, we normalize the collision frequency by the electron mean free path to the surface of the structure. We show that electrochemical gold oxidation provides a stronger effect on collision frequency than these coatings. Similar ellipsometric experiments have previously been conducted to investigate the optical properties of gold oxide, but without taking chemical interface damping into account. The change in reflection from oxidation of gold was solely attributed to the oxide coating. We also show that the chemical interface damping effect saturates at a larger effective oxide thickness, which is attributed to the stabilization of the gold-oxide interface.

The generation and utilization of hot charge carriers in plasmonic materials have emerged as a topic of significant importance, with profound implications across multiple disciplines, including optoelectronics, photovoltaics, photocatalysis, and sensing. In this study, we investigate the hot electron transfer from nanoporous gold (npAu) in dependence of the structure size, utilizing both the nanoscale feature size and the interconnected nature of this material. We employ photoelectron injection from nanoporous gold into the electrolyte under UV illumination as a test electron transfer process. Nanoporous gold thin films with sub-10 nm initial ligament diameter are stepwise coarsened by potential cycles in a photoelectrochemical setup, thereby allowing us to precisely probe the influence of ligament diameter on the photocurrent response. The resulting ligament diameter variations are confirmed by scanning electron microscopy (SEM) analysis. As the ligament diameter increased from 8 to 16 nm, there was a corresponding decrease in quantum efficiency proportional to the inverse ligament diameter squared. Such dependency is expected for electrons excited by surface collisions. For the small ligament diameter of 10 nm we estimate an emission efficiency of excited 6sp electrons as 3.14%, reaching 23% for the surface excited electrons., Comment: 30 pages, 13 figures

English has long been assumed the $\textit{lingua franca}$ of scientific research, and this notion is reflected in the natural language processing (NLP) research involving scientific document representation. In this position piece, we quantitatively show that the literature is largely multilingual and argue that current models and benchmarks should reflect this linguistic diversity. We provide evidence that text-based models fail to create meaningful representations for non-English papers and highlight the negative user-facing impacts of using English-only models non-discriminately across a multilingual domain. We end with suggestions for the NLP community on how to improve performance on non-English documents.

Aiming to enhance the utilization of metric space by the parametric softmax classifier, recent studies suggest replacing it with a non-parametric alternative. Although a non-parametric classifier may provide better metric space utilization, it introduces the challenge of capturing inter-class relationships. A shared characteristic among prior non-parametric classifiers is the static assignment of labels to prototypes during the training, ie, each prototype consistently represents a class throughout the training course. Orthogonal to previous works, we present a simple yet effective method to optimize the category assigned to each prototype (label-to-prototype assignment) during the training. To this aim, we formalize the problem as a two-step optimization objective over network parameters and label-to-prototype assignment mapping. We solve this optimization using a sequential combination of gradient descent and Bipartide matching. We demonstrate the benefits of the proposed approach by conducting experiments on balanced and long-tail classification problems using different backbone network architectures. In particular, our method outperforms its competitors by 1.22\% accuracy on CIFAR-100, and 2.15\% on ImageNet-200 using a metric space dimension half of the size of its competitors. Code: https://github.com/msed-Ebrahimi/DL2PA_CVPR24, Comment: Accepted to CVPR 2024

Background: Image classification can be considered one of the key pillars of medical image analysis. Deep learning (DL) faces challenges that prevent its practical applications despite the remarkable improvement in medical image classification. The data distribution differences can lead to a drop in the efficiency of DL, known as the domain shift problem. Besides, requiring bulk annotated data for model training, the large size of models, and the privacy-preserving of patients are other challenges of using DL in medical image classification. This study presents a strategy that can address the mentioned issues simultaneously. Method: The proposed domain adaptive model based on knowledge distillation can classify images by receiving limited annotated data of different distributions. The designed multiple teachers-meticulous student model trains a student network that tries to solve the challenges by receiving the parameters of several teacher networks. The proposed model was evaluated using six available datasets of different distributions by defining the respiratory motion artefact detection task. Results: The results of extensive experiments using several datasets show the superiority of the proposed model in addressing the domain shift problem and lack of access to bulk annotated data. Besides, the privacy preservation of patients by receiving only the teacher network parameters instead of the original data and consolidating the knowledge of several DL models into a model with almost similar performance are other advantages of the proposed model. Conclusions: The proposed model can pave the way for practical clinical applications of deep classification methods by achieving the mentioned objectives simultaneously.

Let $\Lambda$ be a finite dimensional algebra. In this paper we show that there is a natural bijection between cosilting modules in Mod$\Lambda$ and semibricks in Mod$\Lambda$ satisfying some condition. Also this bijection restricts to a bijection between all semibricks in mod$\Lambda$ and a certain subclass of cosilting modules. These bijections are generalizations of Asai's correspondence [7] between support $\tau^-$-tilting modules and right finite semibricks.

In this paper, we introduce and study the problem of \textit{binary stretch embedding} of edge-weighted graph. This problem is closely related to the well-known \textit{addressing problem} of Graham and Pollak. Addressing problem is the problem of assigning the shortest possible length strings (called ``addresses") over the alphabet $\{0,1,*\}$ to the vertices of an input graph $G$ with the following property. For every pair $u,v$ of vertices, the number of positions in which one of their addresses is $1$, and the other is $0$ is exactly equal to the distance of $u,v$ in graph $G$. When the addresses do not contain the symbol $*$, the problem is called \textit{isometric hypercube embedding}. As far as we know, the isometric hypercube embedding was introduced by Firsov in 1965. It is known that such addresses do not exist for general graphs. Inspired by the addressing problem, in this paper, we introduce the \textit{binary stretch embedding problem}, or BSEP for short, for the edge-weighted undirected graphs. We also argue how this problem is related to other graph embedding problems in the literature. Using tools and techniques such as Hadamard codes and the theory of linear programming, several upper and lower bounds as well as exact solutions for certain classes of graphs will be discovered. As an application of the results in this paper, we derive improved upper bounds or exact values for the maximum size of Lee metric codes of certain parameters.

We develop the fundamentals of nonlinear and anisotropic bending behavior of biomimetic scale plates using a combination of analytical modeling, finite element (FE) computations, and motivational experiments. The analytical architecture-property relationships are derived for both synclastic and anticlastic curvatures. The results show that, as the scales engage, both synclastic and anticlastic deformations show non-linear scale contact kinematics and cross-curvature sensitivity of moments resulting in strong curvature-dependent elastic nonlinearity and emergent anisotropy. The anisotropy of bending rigidities and their evolution with curvature are affected by both the direction and magnitude of bending as well as scale geometry parameters, and their distribution on the substrate. Like earlier beam-like substrates, kinematic locked states were found to occur; however, their existence and evolution are also strongly determined by scale geometry and imposed cross-curvatures. This validated model helps us to quantify bending response, locking behavior, and their geometric dependence, paving the way for a deeper understanding of the nature of nonlinearity and anisotropy of these systems.

A neural network for software compensation was developed for the highly granular CALICE Analogue Hadronic Calorimeter (AHCAL). The neural network uses spatial and temporal event information from the AHCAL and energy information, which is expected to improve sensitivity to shower development and the neutron fraction of the hadron shower. The neural network method produced a depth-dependent energy weighting and a time-dependent threshold for enhancing energy deposits consistent with the timescale of evaporation neutrons. Additionally, it was observed to learn an energy-weighting indicative of longitudinal leakage correction. In addition, the method produced a linear detector response and outperformed a published control method regarding resolution for every particle energy studied.

Recent years have witnessed significant advancement in face recognition (FR) techniques, with their applications widely spread in people's lives and security-sensitive areas. There is a growing need for reliable interpretations of decisions of such systems. Existing studies relying on various mechanisms have investigated the usage of saliency maps as an explanation approach, but suffer from different limitations. This paper first explores the spatial relationship between face image and its deep representation via gradient backpropagation. Then a new explanation approach FGGB has been conceived, which provides precise and insightful similarity and dissimilarity saliency maps to explain the "Accept" and "Reject" decision of an FR system. Extensive visual presentation and quantitative measurement have shown that FGGB achieves superior performance in both similarity and dissimilarity maps when compared to current state-of-the-art explainable face verification approaches.

As applications using immersive media gain increasing attention from both academia and industry, research in the field of point cloud compression has greatly intensified in recent years, leading to the development of the MPEG compression standards V-PCC and G-PCC, as well as the more recent JPEG Pleno learning-based point cloud coding. Each of the above-mentioned standards is based on a different algorithm, introducing distinct types of degradation that may impair the quality of experience when high lossy compression is applied. Although the impact on perceptual quality could be accurately evaluated during subjective quality assessment experiments, objective quality metrics serve as predictors of the visually perceived quality and provide similarity scores without human intervention. Nevertheless, their accuracy can be susceptible to the characteristics of the evaluated media as well as to the type and intensity of the added distortion. While the performance of multiple state-of-the-art objective quality metrics has already been evaluated through their correlation with subjective scores obtained in the presence of artifacts produced by the MPEG standards, no study has evaluated how metrics perform with the more recent JPEG Pleno point cloud coding. In this paper, a study is conducted to benchmark the performance of a large set of objective quality metrics in a subjective dataset including distortions produced by JPEG and MPEG codecs. The dataset also contains three different trade-offs between color and geometry compression for each codec, adding another dimension to the analysis. Performance indexes are computed over the entire dataset but also after splitting according to the codec and to the original model, resulting in detailed insights about the overall performance of each visual quality predictor as well as their cross-content and cross-codec generalization ability.

Pre-trained Large Language Models (LLMs) have significantly advanced natural language processing capabilities but are susceptible to biases present in their training data, leading to unfair outcomes in various applications. While numerous strategies have been proposed to mitigate bias, they often require extensive computational resources and may compromise model performance. In this work, we introduce AXOLOTL, a novel post-processing framework, which operates agnostically across tasks and models, leveraging public APIs to interact with LLMs without direct access to internal parameters. Through a three-step process resembling zero-shot learning, AXOLOTL identifies biases, proposes resolutions, and guides the model to self-debias its outputs. This approach minimizes computational costs and preserves model performance, making AXOLOTL a promising tool for debiasing LLM outputs with broad applicability and ease of use.