Search

The integration of hardware accelerators has significantly advanced the capabilities of modern recommendation systems, enabling the exploration of complex ranking paradigms previously deemed impractical. However, the GPU-based computational costs present substantial challenges. In this paper, we demonstrate our development of an efficiency-driven approach to explore these paradigms, moving beyond traditional reliance on native PyTorch modules. We address the specific challenges posed by ranking models' dependence on categorical features, which vary in length and complicate GPU utilization. We introduce Jagged Feature Interaction Kernels, a novel method designed to extract fine-grained insights from long categorical features through efficient handling of dynamically sized tensors. We further enhance the performance of attention mechanisms by integrating Jagged tensors with Flash Attention. Our novel Jagged Flash Attention achieves up to 9x speedup and 22x memory reduction compared to dense attention. Notably, it also outperforms dense flash attention, with up to 3x speedup and 53% more memory efficiency. In production models, we observe 10% QPS improvement and 18% memory savings, enabling us to scale our recommendation systems with longer features and more complex architectures., Comment: 3 pages, 2 figures

By leveraging the kernel trick in the output space, kernel-induced losses provide a principled way to define structured output prediction tasks for a wide variety of output modalities. In particular, they have been successfully used in the context of surrogate non-parametric regression, where the kernel trick is typically exploited in the input space as well. However, when inputs are images or texts, more expressive models such as deep neural networks seem more suited than non-parametric methods. In this work, we tackle the question of how to train neural networks to solve structured output prediction tasks, while still benefiting from the versatility and relevance of kernel-induced losses. We design a novel family of deep neural architectures, whose last layer predicts in a data-dependent finite-dimensional subspace of the infinite-dimensional output feature space deriving from the kernel-induced loss. This subspace is chosen as the span of the eigenfunctions of a randomly-approximated version of the empirical kernel covariance operator. Interestingly, this approach unlocks the use of gradient descent algorithms (and consequently of any neural architecture) for structured prediction. Experiments on synthetic tasks as well as real-world supervised graph prediction problems show the relevance of our method.

In recommendation systems, high-quality user embeddings can capture subtle preferences, enable precise similarity calculations, and adapt to changing preferences over time to maintain relevance. The effectiveness of recommendation systems depends on the quality of user embedding. We propose to asynchronously learn high fidelity user embeddings for billions of users each day from sequence based multimodal user activities through a Transformer-like large scale feature learning module. The async learned user representations embeddings (ALURE) are further converted to user similarity graphs through graph learning and then combined with user realtime activities to retrieval highly related ads candidates for the ads delivery system. Our method shows significant gains in both offline and online experiments., Comment: Accepted by workshop on Multimodal Representation and Retrieval at SIGIR 2024, Washington DC

Polarons are quasiparticles formed as a result of lattice distortions induced by charge carriers. The single-electron Holstein model captures the fundamentals of single polaron physics. We examine the power of the exponential ansatz for the polaron ground-state wavefunction in its coupled cluster, canonical transformation, and (canonically transformed) perturbative variants across the parameter space of the Holstein model. Our benchmark serves to guide future developments of polaron wavefunctions beyond the single-electron Holstein model., Comment: 10 pages, 6 figures

Safety helmets play a crucial role in protecting workers from head injuries in construction sites, where potential hazards are prevalent. However, currently, there is no approach that can simultaneously achieve both model accuracy and performance in complex environments. In this study, we utilized a Yolo-based model for safety helmet detection, achieved a 2% improvement in mAP (mean Average Precision) performance while reducing parameters and Flops count by over 25%. YOLO(You Only Look Once) is a widely used, high-performance, lightweight model architecture that is well suited for complex environments. We presents a novel approach by incorporating a lightweight feature extraction network backbone based on GhostNetv2, integrating attention modules such as Spatial Channel-wise Attention Net(SCNet) and Coordination Attention Net(CANet), and adopting the Gradient Norm Aware optimizer (GAM) for improved generalization ability. In safety-critical environments, the accurate detection and speed of safety helmets plays a pivotal role in preventing occupational hazards and ensuring compliance with safety protocols. This work addresses the pressing need for robust and efficient helmet detection methods, offering a comprehensive framework that not only enhances accuracy but also improves the adaptability of detection models to real-world conditions. Our experimental results underscore the synergistic effects of GhostNetv2, attention modules, and the GAM optimizer, presenting a compelling solution for safety helmet detection that achieves superior performance in terms of accuracy, generalization, and efficiency.

This paper introduces SparseTSF, a novel, extremely lightweight model for Long-term Time Series Forecasting (LTSF), designed to address the challenges of modeling complex temporal dependencies over extended horizons with minimal computational resources. At the heart of SparseTSF lies the Cross-Period Sparse Forecasting technique, which simplifies the forecasting task by decoupling the periodicity and trend in time series data. This technique involves downsampling the original sequences to focus on cross-period trend prediction, effectively extracting periodic features while minimizing the model's complexity and parameter count. Based on this technique, the SparseTSF model uses fewer than *1k* parameters to achieve competitive or superior performance compared to state-of-the-art models. Furthermore, SparseTSF showcases remarkable generalization capabilities, making it well-suited for scenarios with limited computational resources, small samples, or low-quality data. The code is publicly available at this repository: https://github.com/lss-1138/SparseTSF.

In the rapidly advancing arena of large language models (LLMs), a key challenge is to enhance their capabilities amid a looming shortage of high-quality training data. Our study starts from an empirical strategy for the light continual training of LLMs using their original pre-training data sets, with a specific focus on selective retention of samples that incur moderately high losses. These samples are deemed informative and beneficial for model refinement, contrasting with the highest-loss samples, which would be discarded due to their correlation with data noise and complexity. We then formalize this strategy into a principled framework of Instance-Reweighted Distributionally Robust Optimization (IR-DRO). IR-DRO is designed to dynamically prioritize the training focus on informative samples through an instance reweighting mechanism, streamlined by a closed-form solution for straightforward integration into established training protocols. Through rigorous experimentation with various models and datasets, our findings indicate that our sample-targeted methods significantly improve LLM performance across multiple benchmarks, in both continual pre-training and instruction tuning scenarios. Our codes are available at https://github.com/VITA-Group/HardFocusTraining., Comment: Preprint; updated reference and related works

We propose Any2graph, a generic framework for end-to-end Supervised Graph Prediction (SGP) i.e. a deep learning model that predicts an entire graph for any kind of input. The framework is built on a novel Optimal Transport loss, the Partially-Masked Fused Gromov-Wasserstein, that exhibits all necessary properties (permutation invariance, differentiability and scalability) and is designed to handle any-sized graphs. Numerical experiments showcase the versatility of the approach that outperform existing competitors on a novel challenging synthetic dataset and a variety of real-world tasks such as map construction from satellite image (Sat2Graph) or molecule prediction from fingerprint (Fingerprint2Graph).

Fairness has been a critical issue that affects the adoption of deep learning models in real practice. To improve model fairness, many existing methods have been proposed and evaluated to be effective in their own contexts. However, there is still no systematic evaluation among them for a comprehensive comparison under the same context, which makes it hard to understand the performance distinction among them, hindering the research progress and practical adoption of them. To fill this gap, this paper endeavours to conduct the first large-scale empirical study to comprehensively compare the performance of existing state-of-the-art fairness improving techniques. Specifically, we target the widely-used application scenario of image classification, and utilized three different datasets and five commonly-used performance metrics to assess in total 13 methods from diverse categories. Our findings reveal substantial variations in the performance of each method across different datasets and sensitive attributes, indicating over-fitting on specific datasets by many existing methods. Furthermore, different fairness evaluation metrics, due to their distinct focuses, yield significantly different assessment results. Overall, we observe that pre-processing methods and in-processing methods outperform post-processing methods, with pre-processing methods exhibiting the best performance. Our empirical study offers comprehensive recommendations for enhancing fairness in deep learning models. We approach the problem from multiple dimensions, aiming to provide a uniform evaluation platform and inspire researchers to explore more effective fairness solutions via a set of implications., Comment: Accepted by the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis (ISSTA 2024). Please include ISSTA in any citations

Diffusion-based image synthesis has attracted extensive attention recently. In particular, ControlNet that uses image-based prompts exhibits powerful capability in image tasks such as canny edge detection and generates images well aligned with these prompts. However, vanilla ControlNet generally requires extensive training of around 5000 steps to achieve a desirable control for a single task. Recent context-learning approaches have improved its adaptability, but mainly for edge-based tasks, and rely on paired examples. Thus, two important open issues are yet to be addressed to reach the full potential of ControlNet: (i) zero-shot control for certain tasks and (ii) faster adaptation for non-edge-based tasks. In this paper, we introduce a novel Meta ControlNet method, which adopts the task-agnostic meta learning technique and features a new layer freezing design. Meta ControlNet significantly reduces learning steps to attain control ability from 5000 to 1000. Further, Meta ControlNet exhibits direct zero-shot adaptability in edge-based tasks without any finetuning, and achieves control within only 100 finetuning steps in more complex non-edge tasks such as Human Pose, outperforming all existing methods. The codes is available in https://github.com/JunjieYang97/Meta-ControlNet.

Neural networks have demonstrated success in various domains, yet their performance can be significantly degraded by even a small input perturbation. Consequently, the construction of such perturbations, known as adversarial attacks, has gained significant attention, many of which fall within "white-box" scenarios where we have full access to the neural network. Existing attack algorithms, such as the projected gradient descent (PGD), commonly take the sign function on the raw gradient before updating adversarial inputs, thereby neglecting gradient magnitude information. In this paper, we present a theoretical analysis of how such sign-based update algorithm influences step-wise attack performance, as well as its caveat. We also interpret why previous attempts of directly using raw gradients failed. Based on that, we further propose a new raw gradient descent (RGD) algorithm that eliminates the use of sign. Specifically, we convert the constrained optimization problem into an unconstrained one, by introducing a new hidden variable of non-clipped perturbation that can move beyond the constraint. The effectiveness of the proposed RGD algorithm has been demonstrated extensively in experiments, outperforming PGD and other competitors in various settings, without incurring any additional computational overhead. The codes is available in https://github.com/JunjieYang97/RGD.

Robotic ophthalmic surgery is an emerging technology to facilitate high-precision interventions such as retina penetration in subretinal injection and removal of floating tissues in retinal detachment depending on the input imaging modalities such as microscopy and intraoperative OCT (iOCT). Although iOCT is explored to locate the needle tip within its range-limited ROI, it is still difficult to coordinate iOCT's motion with the needle, especially at the initial target-approaching stage. Meanwhile, due to 2D perspective projection and thus the loss of depth information, current image-based methods cannot effectively estimate the needle tip's trajectory towards both retinal and floating targets. To address this limitation, we propose to use the shadow positions of the target and the instrument tip to estimate their relative depth position and accordingly optimize the instrument tip's insertion trajectory until the tip approaches targets within iOCT's scanning area. Our method succeeds target approaching on a retina model, and achieves an average depth error of 0.0127 mm and 0.3473 mm for floating and retinal targets respectively in the surgical simulator without damaging the retina., Comment: 10 pages

Block2 is an open source framework to implement and perform density matrix renormalization group and matrix product state algorithms. Out-of-the-box it supports the eigenstate, time-dependent, response, and finite-temperature algorithms. In addition, it carries special optimizations for ab initio electronic structure Hamiltonians and implements many quantum chemistry extensions to the density matrix renormalization group, such as dynamical correlation theories. The code is designed with an emphasis on flexibility, extensibility, and efficiency, and to support integration with external numerical packages. Here we explain the design principles and currently supported features and present numerical examples in a range of applications., Comment: 20 pages, 8 figures

Pairwise comparison of graphs is key to many applications in Machine learning ranging from clustering, kernel-based classification/regression and more recently supervised graph prediction. Distances between graphs usually rely on informative representations of these structured objects such as bag of substructures or other graph embeddings. A recently popular solution consists in representing graphs as metric measure spaces, allowing to successfully leverage Optimal Transport, which provides meaningful distances allowing to compare them: the Gromov-Wasserstein distances. However, this family of distances overlooks edge attributes, which are essential for many structured objects. In this work, we introduce an extension of Gromov-Wasserstein distance for comparing graphs whose both nodes and edges have features. We propose novel algorithms for distance and barycenter computation. We empirically show the effectiveness of the novel distance in learning tasks where graphs occur in either input space or output space, such as classification and graph prediction.

Using a systematic ab initio quantum many-body approach that goes beyond low-energy models, we directly compute the superconducting pairing order of several doped cuprate materials and structures. We find that we can correctly capture two well-known trends: the pressure effect, where pairing order increases with intra-layer pressure, and the layer effect, where the pairing order varies with the number of copper-oxygen layers. From these calculations, we observe that the strength of superexchange and the covalency at optimal doping are the best descriptors of the maximal pairing order. Our microscopic analysis further identifies short-range copper spin fluctuations, together with multi-orbital charge fluctuations, as central to the pairing trends. Our work illustrates the possibility of a quantitative computational understanding of unconventional high-temperature superconducting materials., Comment: 10 pages, 5 figures, with supplementary materials

Thermodynamics studies of a prototypical quasi-two-dimensional frustrated magnet Ba$_2$Sn$_2$ZnCr$_{7p}$Ga$_{10-7p}$O$_{22}$ where the magnetic Cr$^{3+}$ ions are arranged in a triangular network of bipyramids show that the magnetic zero-point entropy for $p=0.98$ is 55(1)\% of the entropy expected when the Cr$^{3+}$ moments are fully disordered. Furthermore, when combined with a previous neutron scattering study and the perimeter scaling entropy of a spin jam, the analysis reveals that with decreasing $p$, i.e., doping of the nonmagnetic Ga$^{3+}$ ions, the variation in the magnetic zero-point entropy can be well explained by the combined effects of the zero-point entropy of the spin jam state and that of weakly coupled orphan spins, shedding light on the coexistence of the two types of spin states in quantum magnetism., Comment: 8 pages, 4 figures, and Supplemental Material

Learning to Optimize (L2O) has drawn increasing attention as it often remarkably accelerates the optimization procedure of complex tasks by ``overfitting" specific task type, leading to enhanced performance compared to analytical optimizers. Generally, L2O develops a parameterized optimization method (i.e., ``optimizer") by learning from solving sample problems. This data-driven procedure yields L2O that can efficiently solve problems similar to those seen in training, that is, drawn from the same ``task distribution". However, such learned optimizers often struggle when new test problems come with a substantially deviation from the training task distribution. This paper investigates a potential solution to this open challenge, by meta-training an L2O optimizer that can perform fast test-time self-adaptation to an out-of-distribution task, in only a few steps. We theoretically characterize the generalization of L2O, and further show that our proposed framework (termed as M-L2O) provably facilitates rapid task adaptation by locating well-adapted initial points for the optimizer weight. Empirical observations on several classic tasks like LASSO and Quadratic, demonstrate that M-L2O converges significantly faster than vanilla L2O with only $5$ steps of adaptation, echoing our theoretical results. Codes are available in https://github.com/VITA-Group/M-L2O., Comment: This paper is accepted in ICLR 2023

Learning to optimize (L2O) has gained increasing popularity, which automates the design of optimizers by data-driven approaches. However, current L2O methods often suffer from poor generalization performance in at least two folds: (i) applying the L2O-learned optimizer to unseen optimizees, in terms of lowering their loss function values (optimizer generalization, or ``generalizable learning of optimizers"); and (ii) the test performance of an optimizee (itself as a machine learning model), trained by the optimizer, in terms of the accuracy over unseen data (optimizee generalization, or ``learning to generalize"). While the optimizer generalization has been recently studied, the optimizee generalization (or learning to generalize) has not been rigorously studied in the L2O context, which is the aim of this paper. We first theoretically establish an implicit connection between the local entropy and the Hessian, and hence unify their roles in the handcrafted design of generalizable optimizers as equivalent metrics of the landscape flatness of loss functions. We then propose to incorporate these two metrics as flatness-aware regularizers into the L2O framework in order to meta-train optimizers to learn to generalize, and theoretically show that such generalization ability can be learned during the L2O meta-training process and then transformed to the optimizee loss function. Extensive experiments consistently validate the effectiveness of our proposals with substantially improved generalization on multiple sophisticated L2O models and diverse optimizees. Our code is available at: https://github.com/VITA-Group/Open-L2O/tree/main/Model_Free_L2O/L2O-Entropy., Comment: This paper is accepted in AISTATS 2023

The practical description of disordered chemical reactions, where reactions involve multiple species at multiple sites, is presently challenge using correlated electronic structure methods. Here we describe the gradient theory of multi-fragment density matrix embedding theory, which potentially provides a minimal computational framework to model such processes at the correlated electron level. We present the derivation and implementation of the gradient theory, its validation on model systems and chemical reactions using density matrix embedding, and its application to a molecular dynamics simulation of proton transport in a small water cluster, a simple example of multi-site reaction dynamics.

The development of artificial intelligence (AI) and robotics are both based on the tenet of "science and technology are people-oriented", and both need to achieve efficient communication with the human brain. Based on multi-disciplinary research in systems neuroscience, computer architecture, and functional organic materials, we proposed the concept of using AI to simulate the operating principles and materials of the brain in hardware to develop brain-inspired intelligence technology, and realized the preparation of neuromorphic computing devices and basic materials. We simulated neurons and neural networks in terms of material and morphology, using a variety of organic polymers as the base materials for neuroelectronic devices, for building neural interfaces as well as organic neural devices and silicon neural computational modules. We assemble organic artificial synapses with simulated neurons from silicon-based Field-Programmable Gate Array (FPGA) into organic artificial neurons, the basic components of neural networks, and later construct biological neural network models based on the interpreted neural circuits. Finally, we also discuss how to further build neuromorphic devices based on these organic artificial neurons, which have both a neural interface friendly to nervous tissue and interact with information from real biological neural networks., Comment: This article need to update the corrected figure and data

In the practical application of brain-machine interface technology, the problem often faced is the low information content and high noise of the neural signals collected by the electrode and the difficulty of decoding by the decoder, which makes it difficult for the robotic to obtain stable instructions to complete the task. The idea based on the principle of cooperative shared control can be achieved by extracting general motor commands from brain activity, while the fine details of the movement can be hosted to the robot for completion, or the brain can have complete control. This study proposes a brain-machine interface shared control system based on spiking neural networks for robotic arm movement control and wheeled robots wheel speed control and steering, respectively. The former can reliably control the robotic arm to move to the destination position, while the latter controls the wheeled robots for object tracking and map generation. The results show that the shared control based on brain-inspired intelligence can perform some typical tasks in complex environments and positively improve the fluency and ease of use of brain-machine interaction, and also demonstrate the potential of this control method in clinical applications of brain-machine interfaces., Comment: This article need to update the corrected figure and data