Tong Geng

Yanfei Li, Tong Geng, Samuel Stein et al. · deepmind

Binary neural networks (BNNs) show promising utilization in cost and power-restricted domains such as edge devices and mobile systems. This is due to its significantly less computation and storage demand, but at the cost of degraded performance. To close the accuracy gap, in this paper we propose to add a complementary activation function (AF) ahead of the sign based binarization, and rely on the genetic algorithm (GA) to automatically search for the ideal AFs. These AFs can help extract extra information from the input data in the forward pass, while allowing improved gradient approximation in the backward pass. Fifteen novel AFs are identified through our GA-based search, while most of them show improved performance (up to 2.54% on ImageNet) when testing on different datasets and network models. Our method offers a novel approach for designing general and application-specific BNN architecture. Our code is available at http://github.com/flying-Yan/GAAF.

Hongwu Peng, Shaoyi Huang, Shiyang Chen et al. · deepmind

Transformers are considered one of the most important deep learning models since 2018, in part because it establishes state-of-the-art (SOTA) records and could potentially replace existing Deep Neural Networks (DNNs). Despite the remarkable triumphs, the prolonged turnaround time of Transformer models is a widely recognized roadblock. The variety of sequence lengths imposes additional computing overhead where inputs need to be zero-padded to the maximum sentence length in the batch to accommodate the parallel computing platforms. This paper targets the field-programmable gate array (FPGA) and proposes a coherent sequence length adaptive algorithm-hardware co-design for Transformer acceleration. Particularly, we develop a hardware-friendly sparse attention operator and a length-aware hardware resource scheduling algorithm. The proposed sparse attention operator brings the complexity of attention-based models down to linear complexity and alleviates the off-chip memory traffic. The proposed length-aware resource hardware scheduling algorithm dynamically allocates the hardware resources to fill up the pipeline slots and eliminates bubbles for NLP tasks. Experiments show that our design has very small accuracy loss and has 80.2 $\times$ and 2.6 $\times$ speedup compared to CPU and GPU implementation, and 4 $\times$ higher energy efficiency than state-of-the-art GPU accelerator optimized via CUBLAS GEMM.

Xiaodong Zhao, YiXuan Luo, Juejing Liu et al. · deepmind

Manual analysis of XRD data is usually laborious and time consuming. The deep neural network (DNN) based models trained by synthetic XRD patterns are proved to be an automatic, accurate, and high throughput method to analysis common XRD data collected from solid sample in ambient environment. However, it remains unknown that whether synthetic XRD based models are capable to solve u-XRD mapping data for in-situ experiments involving liquid phase exhibiting lower quality with significant artifacts. In this study, we collected u-XRD mapping data from an LaCl3-calcite hydrothermal fluid system and trained two categories of models to solve the experimental XRD patterns. The models trained by synthetic XRD patterns show low accuracy (as low as 64%) when solving experimental u-XRD mapping data. The accuracy of the DNN models was significantly improved (90% or above) when training them with the dataset containing both synthetic and small number of labeled experimental u-XRD patterns. This study highlighted the importance of labeled experimental patterns on the training of DNN models to solve u-XRD mapping data from in-situ experiments involving liquid phase.

Yuke Wang, Boyuan Feng, Zheng Wang et al. · deepmind

The increasing size of input graphs for graph neural networks (GNNs) highlights the demand for using multi-GPU platforms. However, existing multi-GPU GNN systems optimize the computation and communication individually based on the conventional practice of scaling dense DNNs. For irregularly sparse and fine-grained GNN workloads, such solutions miss the opportunity to jointly schedule/optimize the computation and communication operations for high-performance delivery. To this end, we propose MGG, a novel system design to accelerate full-graph GNNs on multi-GPU platforms. The core of MGG is its novel dynamic software pipeline to facilitate fine-grained computation-communication overlapping within a GPU kernel. Specifically, MGG introduces GNN-tailored pipeline construction and GPU-aware pipeline mapping to facilitate workload balancing and operation overlapping. MGG also incorporates an intelligent runtime design with analytical modeling and optimization heuristics to dynamically improve the execution performance. Extensive evaluation reveals that MGG outperforms state-of-the-art full-graph GNN systems across various settings: on average 4.41X, 4.81X, and 10.83X faster than DGL, MGG-UVM, and ROC, respectively.

Chengming Zhang, Tong Geng, Anqi Guo et al. · deepmind

Graph Neural Networks (GNNs) have drawn tremendous attention due to their unique capability to extend Machine Learning (ML) approaches to applications broadly-defined as having unstructured data, especially graphs. Compared with other Machine Learning (ML) modalities, the acceleration of Graph Neural Networks (GNNs) is more challenging due to the irregularity and heterogeneity derived from graph typologies. Existing efforts, however, have focused mainly on handling graphs' irregularity and have not studied their heterogeneity. To this end we propose H-GCN, a PL (Programmable Logic) and AIE (AI Engine) based hybrid accelerator that leverages the emerging heterogeneity of Xilinx Versal Adaptive Compute Acceleration Platforms (ACAPs) to achieve high-performance GNN inference. In particular, H-GCN partitions each graph into three subgraphs based on its inherent heterogeneity, and processes them using PL and AIE, respectively. To further improve performance, we explore the sparsity support of AIE and develop an efficient density-aware method to automatically map tiles of sparse matrix-matrix multiplication (SpMM) onto the systolic tensor array. Compared with state-of-the-art GCN accelerators, H-GCN achieves, on average, speedups of 1.1~2.3X.

Guangyan Sun, Mingyu Jin, Zhenting Wang et al.

Achieving human-level intelligence requires refining cognitive distinctions between System 1 and System 2 thinking. While contemporary AI, driven by large language models, demonstrates human-like traits, it falls short of genuine cognition. Transitioning from structured benchmarks to real-world scenarios presents challenges for visual agents, often leading to inaccurate and overly confident responses. To address the challenge, we introduce FaST, which incorporates the Fast and Slow Thinking mechanism into visual agents. FaST employs a switch adapter to dynamically select between System 1/2 modes, tailoring the problem-solving approach to different task complexity. It tackles uncertain and unseen objects by adjusting model confidence and integrating new contextual data. With this novel design, we advocate a flexible system, hierarchical reasoning capabilities, and a transparent decision-making pipeline, all of which contribute to its ability to emulate human-like cognitive processes in visual intelligence. Empirical results demonstrate that FaST outperforms various well-known baselines, achieving 80.8% accuracy over VQA^{v2} for visual question answering and 48.7% GIoU score over ReasonSeg for reasoning segmentation, demonstrate FaST's superior performance. Extensive testing validates the efficacy and robustness of FaST's core components, showcasing its potential to advance the development of cognitive visual agents in AI systems. The code is available at ttps://github.com/GuangyanS/Sys2-LLaVA.

Hongwu Peng, Shanglin Zhou, Yukui Luo et al. · deepmind

The rapid growth and deployment of deep learning (DL) has witnessed emerging privacy and security concerns. To mitigate these issues, secure multi-party computation (MPC) has been discussed, to enable the privacy-preserving DL computation. In practice, they often come at very high computation and communication overhead, and potentially prohibit their popularity in large scale systems. Two orthogonal research trends have attracted enormous interests in addressing the energy efficiency in secure deep learning, i.e., overhead reduction of MPC comparison protocol, and hardware acceleration. However, they either achieve a low reduction ratio and suffer from high latency due to limited computation and communication saving, or are power-hungry as existing works mainly focus on general computing platforms such as CPUs and GPUs. In this work, as the first attempt, we develop a systematic framework, PolyMPCNet, of joint overhead reduction of MPC comparison protocol and hardware acceleration, by integrating hardware latency of the cryptographic building block into the DNN loss function to achieve high energy efficiency, accuracy, and security guarantee. Instead of heuristically checking the model sensitivity after a DNN is well-trained (through deleting or dropping some non-polynomial operators), our key design principle is to em enforce exactly what is assumed in the DNN design -- training a DNN that is both hardware efficient and secure, while escaping the local minima and saddle points and maintaining high accuracy. More specifically, we propose a straight through polynomial activation initialization method for cryptographic hardware friendly trainable polynomial activation function to replace the expensive 2P-ReLU operator. We develop a cryptographic hardware scheduler and the corresponding performance model for Field Programmable Gate Arrays (FPGA) platform.

James C. Liang, Yiming Cui, Qifan Wang et al.

This paper presents CLUSTERFORMER, a universal vision model that is based on the CLUSTERing paradigm with TransFORMER. It comprises two novel designs: 1. recurrent cross-attention clustering, which reformulates the cross-attention mechanism in Transformer and enables recursive updates of cluster centers to facilitate strong representation learning; and 2. feature dispatching, which uses the updated cluster centers to redistribute image features through similarity-based metrics, resulting in a transparent pipeline. This elegant design streamlines an explainable and transferable workflow, capable of tackling heterogeneous vision tasks (i.e., image classification, object detection, and image segmentation) with varying levels of clustering granularity (i.e., image-, box-, and pixel-level). Empirical results demonstrate that CLUSTERFORMER outperforms various well-known specialized architectures, achieving 83.41% top-1 acc. over ImageNet-1K for image classification, 54.2% and 47.0% mAP over MS COCO for object detection and instance segmentation, 52.4% mIoU over ADE20K for semantic segmentation, and 55.8% PQ over COCO Panoptic for panoptic segmentation. For its efficacy, we hope our work can catalyze a paradigm shift in universal models in computer vision.

Tong Geng, Chunshu Wu, Yongan Zhang et al.

Graph Convolutional Networks (GCNs) have drawn tremendous attention in the past three years. Compared with other deep learning modalities, high-performance hardware acceleration of GCNs is as critical but even more challenging. The hurdles arise from the poor data locality and redundant computation due to the large size, high sparsity, and irregular non-zero distribution of real-world graphs. In this paper we propose a novel hardware accelerator for GCN inference, called I-GCN, that significantly improves data locality and reduces unnecessary computation. The mechanism is a new online graph restructuring algorithm we refer to as islandization. The proposed algorithm finds clusters of nodes with strong internal but weak external connections. The islandization process yields two major benefits. First, by processing islands rather than individual nodes, there is better on-chip data reuse and fewer off-chip memory accesses. Second, there is less redundant computation as aggregation for common/shared neighbors in an island can be reused. The parallel search, identification, and leverage of graph islands are all handled purely in hardware at runtime working in an incremental pipeline. This is done without any preprocessing of the graph data or adjustment of the GCN model structure. Experimental results show that I-GCN can significantly reduce off-chip accesses and prune 38% of aggregation operations, leading to performance speedups over CPUs, GPUs, the prior art GCN accelerators of 5549x, 403x, and 5.7x on average, respectively.

Hongwu Peng, Ran Ran, Yukui Luo et al.

The growth of Graph Convolution Network (GCN) model sizes has revolutionized numerous applications, surpassing human performance in areas such as personal healthcare and financial systems. The deployment of GCNs in the cloud raises privacy concerns due to potential adversarial attacks on client data. To address security concerns, Privacy-Preserving Machine Learning (PPML) using Homomorphic Encryption (HE) secures sensitive client data. However, it introduces substantial computational overhead in practical applications. To tackle those challenges, we present LinGCN, a framework designed to reduce multiplication depth and optimize the performance of HE based GCN inference. LinGCN is structured around three key elements: (1) A differentiable structural linearization algorithm, complemented by a parameterized discrete indicator function, co-trained with model weights to meet the optimization goal. This strategy promotes fine-grained node-level non-linear location selection, resulting in a model with minimized multiplication depth. (2) A compact node-wise polynomial replacement policy with a second-order trainable activation function, steered towards superior convergence by a two-level distillation approach from an all-ReLU based teacher model. (3) an enhanced HE solution that enables finer-grained operator fusion for node-wise activation functions, further reducing multiplication level consumption in HE-based inference. Our experiments on the NTU-XVIEW skeleton joint dataset reveal that LinGCN excels in latency, accuracy, and scalability for homomorphically encrypted inference, outperforming solutions such as CryptoGCN. Remarkably, LinGCN achieves a 14.2x latency speedup relative to CryptoGCN, while preserving an inference accuracy of 75% and notably reducing multiplication depth.

Xi Xie, Hongwu Peng, Amit Hasan et al.

Graph Convolutional Networks (GCNs) are pivotal in extracting latent information from graph data across various domains, yet their acceleration on mainstream GPUs is challenged by workload imbalance and memory access irregularity. To address these challenges, we present Accel-GCN, a GPU accelerator architecture for GCNs. The design of Accel-GCN encompasses: (i) a lightweight degree sorting stage to group nodes with similar degree; (ii) a block-level partition strategy that dynamically adjusts warp workload sizes, enhancing shared memory locality and workload balance, and reducing metadata overhead compared to designs like GNNAdvisor; (iii) a combined warp strategy that improves memory coalescing and computational parallelism in the column dimension of dense matrices. Utilizing these principles, we formulated a kernel for sparse matrix multiplication (SpMM) in GCNs that employs block-level partitioning and combined warp strategy. This approach augments performance and multi-level memory efficiency and optimizes memory bandwidth by exploiting memory coalescing and alignment. Evaluation of Accel-GCN across 18 benchmark graphs reveals that it outperforms cuSPARSE, GNNAdvisor, and graph-BLAST by factors of 1.17 times, 1.86 times, and 2.94 times respectively. The results underscore Accel-GCN as an effective solution for enhancing GCN computational efficiency.

Hongwu Peng, Deniz Gurevin, Shaoyi Huang et al.

As real-world graphs expand in size, larger GNN models with billions of parameters are deployed. High parameter count in such models makes training and inference on graphs expensive and challenging. To reduce the computational and memory costs of GNNs, optimization methods such as pruning the redundant nodes and edges in input graphs have been commonly adopted. However, model compression, which directly targets the sparsification of model layers, has been mostly limited to traditional Deep Neural Networks (DNNs) used for tasks such as image classification and object detection. In this paper, we utilize two state-of-the-art model compression methods (1) train and prune and (2) sparse training for the sparsification of weight layers in GNNs. We evaluate and compare the efficiency of both methods in terms of accuracy, training sparsity, and training FLOPs on real-world graphs. Our experimental results show that on the ia-email, wiki-talk, and stackoverflow datasets for link prediction, sparse training with much lower training FLOPs achieves a comparable accuracy with the train and prune method. On the brain dataset for node classification, sparse training uses a lower number FLOPs (less than 1/7 FLOPs of train and prune method) and preserves a much better accuracy performance under extreme model sparsity.

Yawen Lu, Qifan Wang, Siqi Ma et al.

Optical flow is an indispensable building block for various important computer vision tasks, including motion estimation, object tracking, and disparity measurement. In this work, we propose TransFlow, a pure transformer architecture for optical flow estimation. Compared to dominant CNN-based methods, TransFlow demonstrates three advantages. First, it provides more accurate correlation and trustworthy matching in flow estimation by utilizing spatial self-attention and cross-attention mechanisms between adjacent frames to effectively capture global dependencies; Second, it recovers more compromised information (e.g., occlusion and motion blur) in flow estimation through long-range temporal association in dynamic scenes; Third, it enables a concise self-learning paradigm and effectively eliminate the complex and laborious multi-stage pre-training procedures. We achieve the state-of-the-art results on the Sintel, KITTI-15, as well as several downstream tasks, including video object detection, interpolation and stabilization. For its efficacy, we hope TransFlow could serve as a flexible baseline for optical flow estimation.

Hongwu Peng, Shanglin Zhou, Yukui Luo et al.

The proliferation of deep learning (DL) has led to the emergence of privacy and security concerns. To address these issues, secure Two-party computation (2PC) has been proposed as a means of enabling privacy-preserving DL computation. However, in practice, 2PC methods often incur high computation and communication overhead, which can impede their use in large-scale systems. To address this challenge, we introduce RRNet, a systematic framework that aims to jointly reduce the overhead of MPC comparison protocols and accelerate computation through hardware acceleration. Our approach integrates the hardware latency of cryptographic building blocks into the DNN loss function, resulting in improved energy efficiency, accuracy, and security guarantees. Furthermore, we propose a cryptographic hardware scheduler and corresponding performance model for Field Programmable Gate Arrays (FPGAs) to further enhance the efficiency of our framework. Experiments show RRNet achieved a much higher ReLU reduction performance than all SOTA works on CIFAR-10 dataset.

Hongwu Peng, Caiwen Ding, Tong Geng et al.

The relentless advancement of artificial intelligence (AI) and machine learning (ML) applications necessitates the development of specialized hardware accelerators capable of handling the increasing complexity and computational demands. Traditional computing architectures, based on the von Neumann model, are being outstripped by the requirements of contemporary AI/ML algorithms, leading to a surge in the creation of accelerators like the Graphcore Intelligence Processing Unit (IPU), Sambanova Reconfigurable Dataflow Unit (RDU), and enhanced GPU platforms. These hardware accelerators are characterized by their innovative data-flow architectures and other design optimizations that promise to deliver superior performance and energy efficiency for AI/ML tasks. This research provides a preliminary evaluation and comparison of these commercial AI/ML accelerators, delving into their hardware and software design features to discern their strengths and unique capabilities. By conducting a series of benchmark evaluations on common DNN operators and other AI/ML workloads, we aim to illuminate the advantages of data-flow architectures over conventional processor designs and offer insights into the performance trade-offs of each platform. The findings from our study will serve as a valuable reference for the design and performance expectations of research prototypes, thereby facilitating the development of next-generation hardware accelerators tailored for the ever-evolving landscape of AI/ML applications. Through this analysis, we aspire to contribute to the broader understanding of current accelerator technologies and to provide guidance for future innovations in the field.

Hao Feng, Boyuan Zhang, Fanjiang Ye et al.

DLRM is a state-of-the-art recommendation system model that has gained widespread adoption across various industry applications. The large size of DLRM models, however, necessitates the use of multiple devices/GPUs for efficient training. A significant bottleneck in this process is the time-consuming all-to-all communication required to collect embedding data from all devices. To mitigate this, we introduce a method that employs error-bounded lossy compression to reduce the communication data size and accelerate DLRM training. We develop a novel error-bounded lossy compression algorithm, informed by an in-depth analysis of embedding data features, to achieve high compression ratios. Moreover, we introduce a dual-level adaptive strategy for error-bound adjustment, spanning both table-wise and iteration-wise aspects, to balance the compression benefits with the potential impacts on accuracy. We further optimize our compressor for PyTorch tensors on GPUs, minimizing compression overhead. Evaluation shows that our method achieves a 1.38$\times$ training speedup with a minimal accuracy impact.

Deniz Gurevin, Mohsin Shan, Tong Geng et al.

In recent years, graph representation learning has gained significant popularity, which aims to generate node embeddings that capture features of graphs. One of the methods to achieve this is employing a technique called random walks that captures node sequences in a graph and then learns embeddings for each node using a natural language processing technique called Word2Vec. These embeddings are then used for deep learning on graph data for classification tasks, such as link prediction or node classification. Prior work operates on pre-collected temporal graph data and is not designed to handle updates on a graph in real-time. Real world graphs change dynamically and their entire temporal updates are not available upfront. In this paper, we propose an end-to-end graph learning pipeline that performs temporal graph construction, creates low-dimensional node embeddings, and trains multi-layer neural network models in an online setting. The training of the neural network models is identified as the main performance bottleneck as it performs repeated matrix operations on many sequentially connected low-dimensional kernels. We propose to unlock fine-grain parallelism in these low-dimensional kernels to boost performance of model training.

Chuan Liu, Chunshu Wu, Shihui Cao et al.

The rapid development of AI highlights the pressing need for sustainable energy, a critical global challenge for decades. Nuclear fusion, generally seen as an ultimate solution, has been the focus of intensive research for nearly a century, with investments reaching hundreds of billions of dollars. Recent advancements in Inertial Confinement Fusion have drawn significant attention to fusion research, in which Laser-Plasma Interaction (LPI) is critical for ensuring fusion stability and efficiency. However, the complexity of LPI upon fusion ignition makes analytical approaches impractical, leaving researchers depending on extremely computation-demanding Particle-in-Cell (PIC) simulations to generate data, presenting a significant bottleneck to advancing fusion research. In response, this work introduces Diff-PIC, a novel framework that leverages conditional diffusion models as a computationally efficient alternative to PIC simulations for generating high-fidelity scientific LPI data. In this work, physical patterns captured by PIC simulations are distilled into diffusion models associated with two tailored enhancements: (1) To effectively capture the complex relationships between physical parameters and corresponding outcomes, the parameters are encoded in a physically-informed manner. (2) To further enhance efficiency while maintaining high fidelity and physical validity, the rectified flow technique is employed to transform our model into a one-step conditional diffusion model. Experimental results show that Diff-PIC achieves 16,200$\times$ speedup compared to traditional PIC on a 100 picosecond simulation, with an average reduction in MAE / RMSE / FID of 59.21% / 57.15% / 39.46% with respect to two other SOTA data generation approaches.

Runjia Zeng, Cheng Han, Qifan Wang et al.

With the scale of vision Transformer-based models continuing to grow, finetuning these large-scale pretrained models for new tasks has become increasingly parameter-intensive. Visual prompt tuning is introduced as a parameter-efficient finetuning (PEFT) method to this trend. Despite its successes, a notable research challenge persists within almost all PEFT approaches: significant performance degradation is observed when there is a substantial disparity between the datasets applied in pretraining and finetuning phases. To address this challenge, we draw inspiration from human visual cognition, and propose the Visual Fourier Prompt Tuning (VFPT) method as a general and effective solution for adapting large-scale transformer-based models. Our approach innovatively incorporates the Fast Fourier Transform into prompt embeddings and harmoniously considers both spatial and frequency domain information. Apart from its inherent simplicity and intuitiveness, VFPT exhibits superior performance across all datasets, offering a general solution to dataset challenges, irrespective of data disparities. Empirical results demonstrate that our approach outperforms current state-of-the-art baselines on two benchmarks, with low parameter usage (e.g., 0.57% of model parameters on VTAB-1k) and notable performance enhancements (e.g., 73.20% of mean accuracy on VTAB-1k). Our code is avaliable at https://github.com/runtsang/VFPT.

Mingkai Chen, Taowen Wang, Shihui Cao et al.

Controlled fusion energy is deemed pivotal for the advancement of human civilization. In this study, we introduce $\textbf{LPI-LLM}$, a novel integration of Large Language Models (LLMs) with classical reservoir computing paradigms tailored to address a critical challenge, Laser-Plasma Instabilities ($\texttt{LPI}$), in Inertial Confinement Fusion ($\texttt{ICF}$). Our approach offers several key contributions: Firstly, we propose the $\textit{LLM-anchored Reservoir}$, augmented with a $\textit{Fusion-specific Prompt}$, enabling accurate forecasting of $\texttt{LPI}$-generated-hot electron dynamics during implosion. Secondly, we develop $\textit{Signal-Digesting Channels}$ to temporally and spatially describe the driver laser intensity across time, capturing the unique characteristics of $\texttt{ICF}$ inputs. Lastly, we design the $\textit{Confidence Scanner}$ to quantify the confidence level in forecasting, providing valuable insights for domain experts to design the $\texttt{ICF}$ process. Extensive experiments demonstrate the superior performance of our method, achieving 1.90 CAE, 0.14 $\texttt{top-1}$ MAE, and 0.11 $\texttt{top-5}$ MAE in predicting Hard X-ray ($\texttt{HXR}$) energies emitted by the hot electrons in $\texttt{ICF}$ implosions, which presents state-of-the-art comparisons against concurrent best systems. Additionally, we present $\textbf{LPI4AI}$, the first $\texttt{LPI}$ benchmark based on physical experiments, aimed at fostering novel ideas in $\texttt{LPI}$ research and enhancing the utility of LLMs in scientific exploration. Overall, our work strives to forge an innovative synergy between AI and $\texttt{ICF}$ for advancing fusion energy.

Chunshu Wu, Ruibing Song, Sushant Kondguli et al.

Real-time image segmentation is a key enabler for AR/VR, robotics, drones, and autonomous systems, where tight accuracy, latency, and energy budgets must be met on resource-constrained edge devices. While U-Net offers a favorable balance of accuracy and efficiency compared to large transformer-based models, achieving real-time performance on high-resolution input remains challenging due to compute, memory, and power limits. Extreme quantization, particularly binary networks, is appealing for its hardware-friendly operations. However, two obstacles limit practicality: (1) severe accuracy degradation, and (2) a lack of end-to-end implementations that deliver efficiency on general-purpose GPUs. We make two empirical observations that guide our design. (1) An explicit zero state is essential: training with zero masking to binary U-Net weights yields noticeable sparsity. (2) Quantization sensitivity is uniform across layers. Motivated by these findings, we introduce Masked Binary U-Net (MBU-Net), obtained through a cost-aware masking strategy that prioritizes masking where it yields the highest accuracy-per-cost, reconciling accuracy with near-binary efficiency. To realize these gains in practice, we develop a GPU execution framework that maps MBU-Net to Tensor Cores via a subtractive bit-encoding scheme, efficiently implementing masked binary weights with binary activations. This design leverages native binary Tensor Core BMMA instructions, enabling high throughput and energy savings on widely available GPUs. Across 3 segmentation benchmarks, MBU-Net attains near full-precision accuracy (3% average drop) while delivering 2.04x speedup and 3.54x energy reductions over a 16-bit floating point U-Net.

Runjia Zeng, Guangyan Sun, Qifan Wang et al.

Considering deep neural networks as manifold mappers, the pretrain-then-fine-tune paradigm can be interpreted as a two-stage process: pretrain establishes a broad knowledge base, and fine-tune adjusts the model parameters to activate specific neural pathways to align with the target manifold. Although prior fine-tuning approaches demonstrate success, their rigid parameter space limits their ability to dynamically activate appropriate neural pathways, rendering them ill-equipped to adapt flexibly to the diverse and evolving data distributions. In light of this view, we propose a novel approach, Mixture of Expert Prompt Tuning (MEPT), as an effective and efficient manifold-mapping framework. MEPT leverages the Mixture of Experts architecture by integrating multiple prompt experts to adaptively learn diverse and non-stationary data distributions. Empirical evaluations demonstrate that MEPT outperforms several state-of-the-art parameter efficient baselines on SuperGLUE, achieving notable improvements in mean accuracy (e.g., 1.94%) while significantly reducing activated prompts by 79.25%. The effectiveness of MEPT is further supported by theoretical insights from manifold learning and validated through neural activation pathway visualization results. Our code is avaliable at https://runjia.tech/emnlp_mept/.

Huawei Lin, Tong Geng, Zhaozhuo Xu et al.

Autoregressive (AR) models have recently shown strong performance in image generation, where a critical component is the visual tokenizer (VT) that maps continuous pixel inputs to discrete token sequences. The quality of the VT largely defines the upper bound of AR model performance. However, current discrete VTs fall significantly behind continuous variational autoencoders (VAEs), leading to degraded image reconstructions and poor preservation of details and text. Existing benchmarks focus on end-to-end generation quality, without isolating VT performance. To address this gap, we introduce VTBench, a comprehensive benchmark that systematically evaluates VTs across three core tasks: Image Reconstruction, Detail Preservation, and Text Preservation, and covers a diverse range of evaluation scenarios. We systematically assess state-of-the-art VTs using a set of metrics to evaluate the quality of reconstructed images. Our findings reveal that continuous VAEs produce superior visual representations compared to discrete VTs, particularly in retaining spatial structure and semantic detail. In contrast, the degraded representations produced by discrete VTs often lead to distorted reconstructions, loss of fine-grained textures, and failures in preserving text and object integrity. Furthermore, we conduct experiments on GPT-4o image generation and discuss its potential AR nature, offering new insights into the role of visual tokenization. We release our benchmark and codebase publicly to support further research and call on the community to develop strong, general-purpose open-source VTs.

Runjia Zeng, James Chenhao Liang, Cheng Han et al.

Training large language models (LLMs) from scratch can yield models with unique functionalities and strengths, but it is costly and often leads to redundant capabilities. A more cost-effective alternative is to fuse existing pre-trained LLMs with different architectures into a more powerful model. However, a key challenge in existing model fusion is their dependence on manually predefined vocabulary alignment, which may not generalize well across diverse contexts, leading to performance degradation in several evaluation. To solve this, we draw inspiration from distribution learning and propose the probabilistic token alignment method as a general and soft mapping for alignment, named as PTA-LLM. Our approach innovatively reformulates token alignment into a classic mathematical problem: optimal transport, seamlessly leveraging distribution-aware learning to facilitate more coherent model fusion. Apart from its inherent generality, PTA-LLM exhibits interpretability from a distributional perspective, offering insights into the essence of the token alignment. Empirical results demonstrate that probabilistic token alignment enhances the target model's performance across multiple capabilities. Our code is avaliable at https://runjia.tech/neurips_pta-llm/.

Haoran You, Tong Geng, Yongan Zhang et al.

Graph Convolutional Networks (GCNs) have emerged as the state-of-the-art graph learning model. However, it can be notoriously challenging to inference GCNs over large graph datasets, limiting their application to large real-world graphs and hindering the exploration of deeper and more sophisticated GCN graphs. This is because real-world graphs can be extremely large and sparse. Furthermore, the node degree of GCNs tends to follow the power-law distribution and therefore have highly irregular adjacency matrices, resulting in prohibitive inefficiencies in both data processing and movement and thus substantially limiting the achievable GCN acceleration efficiency. To this end, this paper proposes a GCN algorithm and accelerator Co-Design framework dubbed GCoD which can largely alleviate the aforementioned GCN irregularity and boost GCNs' inference efficiency. Specifically, on the algorithm level, GCoD integrates a split and conquer GCN training strategy that polarizes the graphs to be either denser or sparser in local neighborhoods without compromising the model accuracy, resulting in graph adjacency matrices that (mostly) have merely two levels of workload and enjoys largely enhanced regularity and thus ease of acceleration. On the hardware level, we further develop a dedicated two-pronged accelerator with a separated engine to process each of the aforementioned denser and sparser workloads, further boosting the overall utilization and acceleration efficiency. Extensive experiments and ablation studies validate that our GCoD consistently reduces the number of off-chip accesses, leading to speedups of 15286x, 294x, 7.8x, and 2.5x as compared to CPUs, GPUs, and prior-art GCN accelerators including HyGCN and AWB-GCN, respectively, while maintaining or even improving the task accuracy. Codes are available at https://github.com/RICE-EIC/GCoD.

Huawei Lin, Yunzhi Shi, Tong Geng et al.

Multimodal large language models (MLLMs) have shown strong capabilities but remain limited to fixed modality pairs and require costly fine-tuning with large aligned datasets. Building fully omni-capable models that can integrate text, images, audio, and video remains impractical and lacks robust reasoning support. In this paper, we propose an Agent-Omni framework that coordinates existing foundation models through a master-agent system, enabling flexible multimodal reasoning without retraining. The master agent interprets user intent, delegates subtasks to modality-specific agents, and integrates their outputs into coherent responses. Extensive experiments across text, image, audio, video, and omni benchmarks show that Agent-Omni consistently achieves state-of-the-art performance, particularly on tasks requiring complex cross-modal reasoning. Its agent-based design enables seamless integration of specialized foundation models, ensuring adaptability to diverse inputs while maintaining transparency and interpretability. In addition, the framework is modular and easily extensible, allowing future improvements as stronger models become available.

Haixi Zhang, Aiyinsi Zuo, Zirui Li et al.

Amidst the rapid advancement of camera-based autonomous driving technology, effectiveness is often prioritized with limited attention to computational efficiency. To address this issue, this paper introduces LRHPerception, a real-time monocular perception package for autonomous driving that uses single-view camera video to interpret the surrounding environment. The proposed system combines the computational efficiency of end-to-end learning with the rich representational detail of local mapping methodologies. With significant improvements in object tracking and prediction, road segmentation, and depth estimation integrated into a unified framework, LRHPerception processes monocular image data into a five-channel tensor consisting of RGB, road segmentation, and pixel-level depth estimation, augmented with object detection and trajectory prediction. Experimental results demonstrate strong performance, achieving real-time processing at 29 FPS on a single GPU, representing a 555% speedup over the fastest mapping-based approach.

Huawei Lin, Yingjie Lao, Tong Geng et al.

Large Language Models (LLMs) are vulnerable to attacks like prompt injection, backdoor attacks, and adversarial attacks, which manipulate prompts or models to generate harmful outputs. In this paper, departing from traditional deep learning attack paradigms, we explore their intrinsic relationship and collectively term them Prompt Trigger Attacks (PTA). This raises a key question: Can we determine if a prompt is benign or poisoned? To address this, we propose UniGuardian, the first unified defense mechanism designed to detect prompt injection, backdoor attacks, and adversarial attacks in LLMs. Additionally, we introduce a single-forward strategy to optimize the detection pipeline, enabling simultaneous attack detection and text generation within a single forward pass. Our experiments confirm that UniGuardian accurately and efficiently identifies malicious prompts in LLMs.

Cheng Han, Yawen Lu, Guohao Sun et al.

In this work, we introduce the Prototypical Transformer (ProtoFormer), a general and unified framework that approaches various motion tasks from a prototype perspective. ProtoFormer seamlessly integrates prototype learning with Transformer by thoughtfully considering motion dynamics, introducing two innovative designs. First, Cross-Attention Prototyping discovers prototypes based on signature motion patterns, providing transparency in understanding motion scenes. Second, Latent Synchronization guides feature representation learning via prototypes, effectively mitigating the problem of motion uncertainty. Empirical results demonstrate that our approach achieves competitive performance on popular motion tasks such as optical flow and scene depth. Furthermore, it exhibits generality across various downstream tasks, including object tracking and video stabilization.

Yanfei Li, Juejing Liu, Xiaodong Zhao et al.

Traditional analysis of highly distorted micro-X-ray diffraction (μ-XRD) patterns from hydrothermal fluid environments is a time-consuming process, often requiring substantial data preprocessing and labeled experimental data. This study demonstrates the potential of deep learning with a multitask learning (MTL) architecture to overcome these limitations. We trained MTL models to identify phase information in μ-XRD patterns, minimizing the need for labeled experimental data and masking preprocessing steps. Notably, MTL models showed superior accuracy compared to binary classification CNNs. Additionally, introducing a tailored cross-entropy loss function improved MTL model performance. Most significantly, MTL models tuned to analyze raw and unmasked XRD patterns achieved close performance to models analyzing preprocessed data, with minimal accuracy differences. This work indicates that advanced deep learning architectures like MTL can automate arduous data handling tasks, streamline the analysis of distorted XRD patterns, and reduce the reliance on labor-intensive experimental datasets.

Yongan Zhang, Haoran You, Yonggan Fu et al.

Graph Neural Networks (GNNs) have emerged as the state-of-the-art (SOTA) method for graph-based learning tasks. However, it still remains prohibitively challenging to inference GNNs over large graph datasets, limiting their application to large-scale real-world tasks. While end-to-end jointly optimizing GNNs and their accelerators is promising in boosting GNNs' inference efficiency and expediting the design process, it is still underexplored due to the vast and distinct design spaces of GNNs and their accelerators. In this work, we propose G-CoS, a GNN and accelerator co-search framework that can automatically search for matched GNN structures and accelerators to maximize both task accuracy and acceleration efficiency. Specifically, GCoS integrates two major enabling components: (1) a generic GNN accelerator search space which is applicable to various GNN structures and (2) a one-shot GNN and accelerator co-search algorithm that enables simultaneous and efficient search for optimal GNN structures and their matched accelerators. To the best of our knowledge, G-CoS is the first co-search framework for GNNs and their accelerators. Extensive experiments and ablation studies show that the GNNs and accelerators generated by G-CoS consistently outperform SOTA GNNs and GNN accelerators in terms of both task accuracy and hardware efficiency, while only requiring a few hours for the end-to-end generation of the best matched GNNs and their accelerators.

Hongwu Peng, Shanglin Zhou, Scott Weitze et al.

Being able to learn from complex data with phase information is imperative for many signal processing applications. Today' s real-valued deep neural networks (DNNs) have shown efficiency in latent information analysis but fall short when applied to the complex domain. Deep complex networks (DCN), in contrast, can learn from complex data, but have high computational costs; therefore, they cannot satisfy the instant decision-making requirements of many deployable systems dealing with short observations or short signal bursts. Recent, Binarized Complex Neural Network (BCNN), which integrates DCNs with binarized neural networks (BNN), shows great potential in classifying complex data in real-time. In this paper, we propose a structural pruning based accelerator of BCNN, which is able to provide more than 5000 frames/s inference throughput on edge devices. The high performance comes from both the algorithm and hardware sides. On the algorithm side, we conduct structural pruning to the original BCNN models and obtain 20 $\times$ pruning rates with negligible accuracy loss; on the hardware side, we propose a novel 2D convolution operation accelerator for the binary complex neural network. Experimental results show that the proposed design works with over 90% utilization and is able to achieve the inference throughput of 5882 frames/s and 4938 frames/s for complex NIN-Net and ResNet-18 using CIFAR-10 dataset and Alveo U280 Board.

Boyuan Feng, Yuke Wang, Tong Geng et al.

Over the years, accelerating neural networks with quantization has been widely studied. Unfortunately, prior efforts with diverse precisions (e.g., 1-bit weights and 2-bit activations) are usually restricted by limited precision support on GPUs (e.g., int1 and int4). To break such restrictions, we introduce the first Arbitrary Precision Neural Network framework (APNN-TC) to fully exploit quantization benefits on Ampere GPU Tensor Cores. Specifically, APNN-TC first incorporates a novel emulation algorithm to support arbitrary short bit-width computation with int1 compute primitives and XOR/AND Boolean operations. Second, APNN-TC integrates arbitrary precision layer designs to efficiently map our emulation algorithm to Tensor Cores with novel batching strategies and specialized memory organization. Third, APNN-TC embodies a novel arbitrary precision NN design to minimize memory access across layers and further improve performance. Extensive evaluations show that APNN-TC can achieve significant speedup over CUTLASS kernels and various NN models, such as ResNet and VGG.

Yanfei Li, Tong Geng, Ang Li et al.

Binarized neural networks, or BNNs, show great promise in edge-side applications with resource limited hardware, but raise the concerns of reduced accuracy. Motivated by the complex neural networks, in this paper we introduce complex representation into the BNNs and propose Binary complex neural network -- a novel network design that processes binary complex inputs and weights through complex convolution, but still can harvest the extraordinary computation efficiency of BNNs. To ensure fast convergence rate, we propose novel BCNN based batch normalization function and weight initialization function. Experimental results on Cifar10 and ImageNet using state-of-the-art network models (e.g., ResNet, ResNetE and NIN) show that BCNN can achieve better accuracy compared to the original BNN models. BCNN improves BNN by strengthening its learning capability through complex representation and extending its applicability to complex-valued input data. The source code of BCNN will be released on GitHub.

Tong Geng, Xiliang Lin, Harikesh S. Nair et al.

Comparison Lift is an experimentation-as-a-service (EaaS) application for testing online advertising audiences and creatives at JD.com. Unlike many other EaaS tools that focus primarily on fixed sample A/B testing, Comparison Lift deploys a custom bandit-based experimentation algorithm. The advantages of the bandit-based approach are two-fold. First, it aligns the randomization induced in the test with the advertiser's goals from testing. Second, by adapting experimental design to information acquired during the test, it reduces substantially the cost of experimentation to the advertiser. Since launch in May 2019, Comparison Lift has been utilized in over 1,500 experiments. We estimate that utilization of the product has helped increase click-through rates of participating advertising campaigns by 46% on average. We estimate that the adaptive design in the product has generated 27% more clicks on average during testing compared to a fixed sample A/B design. Both suggest significant value generation and cost savings to advertisers from the product.

Tong Geng, Ang Li, Runbin Shi et al.

Deep learning systems have been successfully applied to Euclidean data such as images, video, and audio. In many applications, however, information and their relationships are better expressed with graphs. Graph Convolutional Networks (GCNs) appear to be a promising approach to efficiently learn from graph data structures, having shown advantages in many critical applications. As with other deep learning modalities, hardware acceleration is critical. The challenge is that real-world graphs are often extremely large and unbalanced; this poses significant performance demands and design challenges. In this paper, we propose Autotuning-Workload-Balancing GCN (AWB-GCN) to accelerate GCN inference. To address the issue of workload imbalance in processing real-world graphs, three hardware-based autotuning techniques are proposed: dynamic distribution smoothing, remote switching, and row remapping. In particular, AWB-GCN continuously monitors the sparse graph pattern, dynamically adjusts the workload distribution among a large number of processing elements (up to 4K PEs), and, after converging, reuses the ideal configuration. Evaluation is performed using an Intel D5005 FPGA with five commonly-used datasets. Results show that 4K-PE AWB-GCN can significantly elevate PE utilization by 7.7x on average and demonstrate considerable performance speedups over CPUs (3255x), GPUs (80.3x), and a prior GCN accelerator (5.1x).

Tong Geng, Xiliang Lin, Harikesh S. Nair

Firms implementing digital advertising campaigns face a complex problem in determining the right match between their advertising creatives and target audiences. Typical solutions to the problem have leveraged non-experimental methods, or used "split-testing" strategies that have not explicitly addressed the complexities induced by targeted audiences that can potentially overlap with one another. This paper presents an adaptive algorithm that addresses the problem via online experimentation. The algorithm is set up as a contextual bandit and addresses the overlap issue by partitioning the target audiences into disjoint, non-overlapping sub-populations. It learns an optimal creative display policy in the disjoint space, while assessing in parallel which creative has the best match in the space of possibly overlapping target audiences. Experiments show that the proposed method is more efficient compared to naive "split-testing" or non-adaptive "A/B/n" testing based methods. We also describe a testing product we built that uses the algorithm. The product is currently deployed on the advertising platform of JD.com, an eCommerce company and a publisher of digital ads in China.

Tong Geng, Tianqi Wang, Ang Li et al.

Deep Neural Networks (DNNs) have revolutionized numerous applications, but the demand for ever more performance remains unabated. Scaling DNN computations to larger clusters is generally done by distributing tasks in batch mode using methods such as distributed synchronous SGD. Among the issues with this approach is that to make the distributed cluster work with high utilization, the workload distributed to each node must be large, which implies nontrivial growth in the SGD mini-batch size. In this paper, we propose a framework called FPDeep, which uses a hybrid of model and layer parallelism to configure distributed reconfigurable clusters to train DNNs. This approach has numerous benefits. First, the design does not suffer from batch size growth. Second, novel workload and weight partitioning leads to balanced loads of both among nodes. And third, the entire system is a fine-grained pipeline. This leads to high parallelism and utilization and also minimizes the time features need to be cached while waiting for back-propagation. As a result, storage demand is reduced to the point where only on-chip memory is used for the convolution layers. We evaluate FPDeep with the Alexnet, VGG-16, and VGG-19 benchmarks. Experimental results show that FPDeep has good scalability to a large number of FPGAs, with the limiting factor being the FPGA-to-FPGA bandwidth. With 6 transceivers per FPGA, FPDeep shows linearity up to 83 FPGAs. Energy efficiency is evaluated with respect to GOPs/J. FPDeep provides, on average, 6.36x higher energy efficiency than comparable GPU servers.