Large Language Models (LLMs) pose a new paradigm of modeling and computation for information tasks. Recommendation systems are a critical application domain poised to benefit significantly from the sequence modeling capabilities and world knowledge inherent in these large models. In this paper, we introduce PLUM, a framework designed to adapt pre-trained LLMs for industry-scale recommendation tasks. PLUM consists of item tokenization using Semantic IDs, continued pre-training (CPT) on domain-specific data, and task-specific fine-tuning for recommendation objectives. For fine-tuning, we focus particularly on generative retrieval, where the model is directly trained to generate Semantic IDs of recommended items based on user context. We conduct comprehensive experiments on large-scale internal video recommendation datasets. Our results demonstrate that PLUM achieves substantial improvements for retrieval compared to a heavily-optimized production model built with large embedding tables. We also present a scaling study for the model's retrieval performance, our learnings about CPT, a few enhancements to Semantic IDs, along with an overview of the training and inference methods that enable launching this framework to billions of users in YouTube.

The semantic reconstruction of a scene relies in part on the curvilinear structure inherent in images. The recovery of curvilinear structure is not only key to the representation of objects via ridges and other object curves but is also critical to the reconstruction from texture-poor images which lack a sufficient number of features. Prior methods advocate for the recovery of curve segments from images and reconstructing these into an organized collection of 3D curve segments often referred to as the 3D curve sketch, which serves as the basis for further reconstruction of curves and surfaces. Observing that the process of edge grouping can lead to fictitious curves or missing veridical groupings, this paper advocates for a reconstruction of curvilinear structure directly from image edges in the form of a 3D edge sketch. The multiview reconstruction of edges faces significant combinatorial challenges which are effectively addressed in this paper. We demonstrate through experiments that the 3D edge sketch recovers a vast majority of the curvilinear structure and is a reliable substrate from which 3D curves can be constructed.

In many applications, we regularly face the fundamental problem of allocating a common resource (funding, time, energy, etc.) among a network of processes that evolve in a continuous-space according to a birth-death dynamics. The state of each process tends to gradually improve with the resource and gradually degrade without it. Formulated as a Constrained Markov Decision Process (CMDP), the problem is typically attacked in the continuous space directly using the function approximation method or in discrete space with a fixed granularity. In this work, we investigated an alternative method based on the fact that the granularity of discretization has a crucial impact on the size and the evolution of the state-space. Increasing the granularity has the desirable effect of increasing the control of the processes (due to increased interaction regularity), but it also comes with the burden of an increasing state-space. Without function approximation, it is well-known that finding the optimal solution of CMDP is formidably difficult as the state-space grows. We have taken a fresh look at designing State-Independent policies whose complexity does not scale with the discretization granularity parameternof the underlying continuous space processes. In particular, for a constrained-resource allocation problem over birth-death type processes, we developed state-independent policies that guarantee asymptotic-optimality as the discretization granularityngrows. We also show, through numerical comparisons, that our design has a lower running time compared to alternative designs including index-based methods and function approximation methods.

Randomly-hashed item ids are used ubiquitously in recommendation models. However, the learned representations from random hashing prevents generalization across similar items, causing problems of learning unseen and long-tail items, especially when item corpus is large, power-law distributed, and evolving dynamically. In this paper, we propose using content-derived features as a replacement for random ids. We show that simply replacing ID features with content-based embeddings can cause a drop in quality due to reduced memorization capability. To strike a good balance of memorization and generalization, we propose to use Semantic IDs [15], a compact and discrete item representation, as a replacement for random item ids. Semantic IDs are learned from frozen content embeddings using RQ-VAE and thus can capture the hierarchy of concepts in items. Similar to content embeddings, the compactness of Semantic IDs poses a problem of adaption in recommendation models. We propose novel methods for adapting Semantic IDs in industry-scale ranking models, through hashing sub-pieces of of the Semantic-ID sequences. In particular, we find that the SentencePiece model [10] that is commonly used in LLM tokenization outperforms manually crafted pieces such as N-grams. To the end, we evaluate our approaches in a real-world ranking model for YouTube recommendations. Our experiments demonstrate that Semantic IDs can replace the direct use of video IDs by improving the generalization ability on new and long-tail item slices without sacrificing overall model quality.

Local Climate Zone (LCZ) classification from remote sensing images plays a crucial role in quantifying the urban heat island effect. However, the performance of LCZ classification has not been satisfactory so far, especially for built-up area categories. To alleviate this issue, we introduce a novel network architecture, DS-LCZNET, which incorporates a Dynamic Spatial Feature Enhancement (DSFE) module for capturing complex spatial information and a SAR-MS Fusion (SMF) module to improve feature integration from SAR and MS data. Extensive experiments demonstrate that DS-LCZNET significantly enhances classification performance, achieving a 3.55% increase in overall accuracy, a 1.18% improvement in average accuracy (AA), and a 3.88% rise in the kappa (x100) coefficient compared to the current leading baseline, MsF- LCZ-Net. The codes will be publicly available at: https://github.com/zhyilin97/DSLCZNET.

In real-world networks, we regularly face the effect of propagating failures over networks, for example, rumors spread over social networks, outages spread over power networks, viruses spread over communication and biological networks. Often, these failures spread over a network of agents with unknown and potentially diverse degrees of vulnerabilities to the propagating phenomenon. In this work, we consider a general network model subject to propagating failures and develop provably fast mechanisms for learning the unknown vulnerabilities of the network with minimal cost incurred in the process. We propose an extension to the classic Independent Cascade (IC) model where we incorporate both node and edge failures with non-uniform costs. From an online learning perspective, the goal is to find an optimal policy to control where to start failures and generate samples. Therefore, we formulate a cost minimization problem with Probably-Approximately-Correct (PAC) type guarantees. As a theoretical benchmark, we design a linear programming problem using a proposed joint Bernstein inequality. Then we characterize the performance of randomized policies that use a fixed budget distribution independent of sampling history. Finally, we propose a fast Lyapunov-based online learning policy, for which we give a formal theoretical analysis. The performance of the policy are validated under extensive numerical studies for both synthetic and real-world networks.

Anomaly detection in computer vision seeks to identify samples outside of a predefined distribution, including texture defect detection and semantic anomaly detection. However, existing methods are difficult to simultaneously achieve high performance for both types of anomaly detection. To address this issue, we propose a new flow-based anomaly detection method. Firstly, we use semantic features extracted from a pre-trained backbone to learn the distribution of normal data from a semantic perspective. Secondly, we introduce a multi-frequency feature fusion module to aggregate semantic and texture information, which substantially improves performance for both types of anomaly detection at the same time. Extensive experiments on multiple well-known datasets demonstrate that our proposed method performs well in both types of anomaly detection, specially, achieves state-of-the-art performance in one-class anomaly detection. The codes will be available at https://github.com/SYLan2019/FOAD-MFFF.

Ship detection in synthetic aperture radar (SAR) images is a major issue in maritime surveillance and port management. Existing challenges are mainly as follows: (1) Tiny ships are mixed with scattered noise spots on the sea. (2) Ships are present in extreme aspect-ratios and various scales. (3) The land background blurs the outline of coastal ships. To address these problems, we propose an efficient detection neural network (DLAHSD) that integrates the Multi-scale Feature Location Fusion (MFLF) module and the Auxiliary Detection Head (ADH) based CenterNet. In addition, we designed a Dynamic Elliptic Gaussian (DEG) module to label the heatmap of ships. Experimental results on the challenging SSDD dataset show that our model offers improved performance over the baseline methods. The codes will be available at https://github.com/SYLan2019/DLAHSD.

In many applications such as statistical machine learning, reinforcement learning, and optimization for large data centers, the increasing data size and model complexity have made it impractical to run optimizations over a single machine. Therefore, solving the distributed optimization problem has become an important task. In this work, we consider a distributed processing network $G = \left( {\mathcal{V},\mathcal{E}} \right)$ with n nodes, where each node i can only evaluate the values of a local function (i.e., has zeroth-order information) and can only communicate with its neighbors. The objective is to reach consensus on the global optimizer of ${\max _{x \in \mathcal{X}}}\frac{1}{n}\sum\nolimits_{i = 1}^n {{f_i}(x)} $. Previous methods either assume first-order gradient information which is not suitable for many model-free learning scenarios, or consider the zeroth-order information but assume convexity of the objective functions and can only guarantee convergence to a stationary point for nonconvex objectives. To address these limitations, we drop both the known gradient assumption and convexity assumption. Instead, we propose a distributed Bayesian framework for the problem with only zeroth-order information and general nonconvex objective functions in a Matérn Reproducing Kernel Hilbert Space (RKHS). Under this framework, we propose an algorithm and show that with high probability it reaches consensus on all nodes and has a sublinear regret with regard to the global optimal. The results are validated under numerical studies.

Unsupervised Domain Adaptation(UDA) in semantic segmentation reduces the dependence on pixel level labeling. It uses labeled source domain datasets and unlabeled target domain images to learn to segment the network. This article proposes a domain adaptive framework that combines decoupled adversarial network and self-training. The problem of over fitting the source domain in domain adaptation and the inability of the network to focus on segmentation tasks has been solved. Considering the impact of the long tailed distribution of data, the Rare Class Sampling (RCS) module is introduced. In order to make full use of pseudo-labels, we designed a self-training UDA scheme using weighted pseudo-labels. At the same time, the RCS module for rare class sampling improves the quality of pseudo-labels by reducing the recognition bias of self-training on public classes. The LoveDA dataset is the latest domain adaptive dataset for land cover mapping. In urban-to-rural and rural-to-urban scenarios, our proposed UDA method has significant advantages.

In a wide variety of applications including online advertising, contractual hiring, and wireless scheduling, the controller is constrained by a stringent budget constraint on the available resources, which are consumed in a random amount by each action, and a stochastic feasibility constraint that may impose important operational limitations on decision-making. In this work, we consider a general model to address such problems, where each action returns a random reward, cost, and penalty from an unknown joint distribution, and the decision-maker aims to maximize the total reward under a budget constraint B on the total cost and a stochastic constraint on the time-average penalty. We propose a novel low-complexity algorithm based on Lyapunov optimization methodology, named LyOn, and prove that for K arms it achieves square root of KBlog(B) regret and zero constraint-violation when B is sufficiently large. The low computational cost and sharp performance bounds of LyOn suggest that Lyapunov-based algorithm design methodology can be effective in solving constrained bandit optimization problems.

Abundant natural resources are the basis of urbanisation and industrialisation. Citizens are the key factor in promoting a sustainable supply of natural resources and the high-quality development of urban areas. This study focuses on the co-production behaviours of citizens regarding urban natural resource assets in the age of big data, and uses the latent Dirichlet allocation algorithm and the stepwise regression analysis method to evaluate citizens’ experiences and feelings related to the urban capitalisation of natural resources. Results show that, firstly, the machine learning algorithm based on natural language processing can effectively identify and deal with the demands of urban natural resource assets. Secondly, in the experience of urban natural resources, citizens pay more attention to the combination of history, culture, infrastructure and natural landscape. Unique natural resource can enhance citizens’ sense of participation. Finally, the scenery, entertainment and quality and value of urban natural resources are the influencing factors of citizens’ satisfaction.

Current VR systems provide various text input methods that enable users to enter text efficiently with virtual keyboards. However, little attention has been paid to facilitate text revision during the VR text input process. We first summarized existing text revision solutions in current VR text input research and found that backspace is the only tool available for text revision with virtual keyboards with few mentioning designs for caret control. To systematically explore VR text revision designs, we presented a design space for VR text revision based on backspace and caret. With the proposed design space, we further analyzed the feasibility of the combined usage of backspace and caret by proposing and evaluating four VR text revision techniques. Outcomes of this research can provide a fundamental understanding of VR text revision solutions (with backspace and caret) and a comparable basis for evaluating future VR text revision techniques.

In this paper, we study meta learning for support (i.e., the set of non-zero entries) recovery in high-dimensional precision matrix estimation where we reduce the sufficient sample complexity in a novel task with the information learned from other auxiliary tasks. In our setup, each task has a different random true precision matrix, each with a possibly different support. We assume that the union of the supports of all the true precision matrices (i.e., the true support union) is small in size. We propose to pool all the samples from different tasks, and \emph{improperly} estimate a single precision matrix by minimizing the $\ell_1$-regularized log-determinant Bregman divergence. We show that with high probability, the support of the \emph{improperly} estimated single precision matrix is equal to the true support union, provided a sufficient number of samples per task $n \in O((\log N)/K)$, for $N$-dimensional vectors and $K$ tasks. That is, one requires less samples per task when more tasks are available. We prove a matching information-theoretic lower bound for the necessary number of samples, which is $n \in \Omega((\log N)/K)$, and thus, our algorithm is minimax optimal. Then for the novel task, we prove that the minimization of the $\ell_1$-regularized log-determinant Bregman divergence with the additional constraint that the support is a subset of the estimated support union could reduce the sufficient sample complexity of successful support recovery to $O(\log(|S_{\text{off}}|))$ where $|S_{\text{off}}|$ is the number of off-diagonal elements in the support union and is much less than $N$ for sparse matrices. We also prove a matching information-theoretic lower bound of $\Omega(\log(|S_{\text{off}}|))$ for the necessary number of samples.