Search

Large language models (LLMs) have demonstrated remarkable capabilities, but their outputs can sometimes be unreliable or factually incorrect. To address this, we introduce Self Logits Evolution Decoding (SLED), a novel decoding framework that enhances the truthfulness of LLMs without relying on external knowledge bases or requiring further fine-tuning. From an optimization perspective, our SLED framework leverages the latent knowledge embedded within the LLM by contrasting the output logits from the final layer with those from early layers. It then utilizes an approximate gradient approach to enable latent knowledge to guide the self-refinement of outputs, thereby effectively improving factual accuracy. Extensive experiments have been conducted on established benchmarks across a diverse range of model families (LLaMA 2, LLaMA 3, Gemma) and scales (from 2B to 70B), including more advanced architectural configurations such as the mixture of experts (MoE). Our evaluation spans a wide variety of tasks, including multi-choice, open-generation, and adaptations to chain-of-thought reasoning tasks. The results demonstrate that SLED consistently improves factual accuracy by up to 20\% compared to existing decoding methods while maintaining natural language fluency and negligible latency overhead. Furthermore, it can be flexibly combined with other decoding methods to further enhance their performance.

The rapid development of large language models (LLMs) has significantly transformed the field of artificial intelligence, demonstrating remarkable capabilities in natural language processing and moving towards multi-modal functionality. These models are increasingly integrated into diverse applications, impacting both research and industry. However, their development and deployment present substantial challenges, including the need for extensive computational resources, high energy consumption, and complex software optimizations. Unlike traditional deep learning systems, LLMs require unique optimization strategies for training and inference, focusing on system-level efficiency. This paper surveys hardware and software co-design approaches specifically tailored to address the unique characteristics and constraints of large language models. This survey analyzes the challenges and impacts of LLMs on hardware and algorithm research, exploring algorithm optimization, hardware design, and system-level innovations. It aims to provide a comprehensive understanding of the trade-offs and considerations in LLM-centric computing systems, guiding future advancements in AI. Finally, we summarize the existing efforts in this space and outline future directions toward realizing production-grade co-design methodologies for the next generation of large language models and AI systems., Comment: Accepted by IEEE Circuits and Systems Magazine

Federated Learning (FL) provides a strong privacy guarantee by enabling local training across edge devices without training data sharing, and Federated Adversarial Training (FAT) further enhances the robustness against adversarial examples, promoting a step toward trustworthy artificial intelligence. However, FAT requires a large model to preserve high accuracy while achieving strong robustness, and it is impractically slow when directly training with memory-constrained edge devices due to the memory-swapping latency. Moreover, existing memory-efficient FL methods suffer from poor accuracy and weak robustness in FAT because of inconsistent local and global models, i.e., objective inconsistency. In this paper, we propose FedProphet, a novel FAT framework that can achieve memory efficiency, adversarial robustness, and objective consistency simultaneously. FedProphet partitions the large model into small cascaded modules such that the memory-constrained devices can conduct adversarial training module-by-module. A strong convexity regularization is derived to theoretically guarantee the robustness of the whole model, and we show that the strong robustness implies low objective inconsistency in FedProphet. We also develop a training coordinator on the server of FL, with Adaptive Perturbation Adjustment for utility-robustness balance and Differentiated Module Assignment for objective inconsistency mitigation. FedProphet empirically shows a significant improvement in both accuracy and robustness compared to previous memory-efficient methods, achieving almost the same performance of end-to-end FAT with 80% memory reduction and up to 10.8x speedup in training time., Comment: Preprint

Previous studies on federated learning (FL) often encounter performance degradation due to data heterogeneity among different clients. In light of the recent advances in multimodal large language models (MLLMs), such as GPT-4v and LLaVA, which demonstrate their exceptional proficiency in multimodal tasks, such as image captioning and multimodal question answering. We introduce a novel federated learning framework, named Multimodal Large Language Model Assisted Federated Learning (MLLM-FL), which which employs powerful MLLMs at the server end to address the heterogeneous and long-tailed challenges. Owing to the advanced cross-modality representation capabilities and the extensive open-vocabulary prior knowledge of MLLMs, our framework is adept at harnessing the extensive, yet previously underexploited, open-source data accessible from websites and powerful server-side computational resources. Hence, the MLLM-FL not only enhances the performance but also avoids increasing the risk of privacy leakage and the computational burden on local devices, distinguishing it from prior methodologies. Our framework has three key stages. Initially, prior to local training on local datasets of clients, we conduct global visual-text pretraining of the model. This pretraining is facilitated by utilizing the extensive open-source data available online, with the assistance of multimodal large language models. Subsequently, the pretrained model is distributed among various clients for local training. Finally, once the locally trained models are transmitted back to the server, a global alignment is carried out under the supervision of MLLMs to further enhance the performance. Experimental evaluations on established benchmarks, show that our framework delivers promising performance in the typical scenarios with data heterogeneity and long-tail distribution across different clients in FL.

In the realm of electronic and electrical engineering, automation of analog circuit is increasingly vital given the complexity and customized requirements of modern applications. However, existing methods only develop search-based algorithms that require many simulation iterations to design a custom circuit topology, which is usually a time-consuming process. To this end, we introduce LaMAGIC, a pioneering language model-based topology generation model that leverages supervised finetuning for automated analog circuit design. LaMAGIC can efficiently generate an optimized circuit design from the custom specification in a single pass. Our approach involves a meticulous development and analysis of various input and output formulations for circuit. These formulations can ensure canonical representations of circuits and align with the autoregressive nature of LMs to effectively addressing the challenges of representing analog circuits as graphs. The experimental results show that LaMAGIC achieves a success rate of up to 96\% under a strict tolerance of 0.01. We also examine the scalability and adaptability of LaMAGIC, specifically testing its performance on more complex circuits. Our findings reveal the enhanced effectiveness of our adjacency matrix-based circuit formulation with floating-point input, suggesting its suitability for handling intricate circuit designs. This research not only demonstrates the potential of language models in graph generation, but also builds a foundational framework for future explorations in automated analog circuit design., Comment: Proceedings of the 41st International Conference on Machine Learning, PMLR 235:6253-6262 https://proceedings.mlr.press/v235/chang24c.html

Despite significant advancements in AI driven by neural networks, tree-based machine learning (TBML) models excel on tabular data. These models exhibit promising energy efficiency, and high performance, particularly when accelerated on analog content-addressable memory (aCAM) arrays. However, optimizing their hardware deployment, especially in leveraging TBML model structure and aCAM circuitry, remains challenging. In this paper, we introduce MonoSparse-CAM, a novel content-addressable memory (CAM) based computing optimization technique. MonoSparse-CAM efficiently leverages TBML model sparsity and CAM array circuits, enhancing processing performance. Our experiments show that MonoSparse-CAM reduces energy consumption by up to 28.56x compared to raw processing and 18.51x compared to existing deployment optimization techniques. Additionally, it consistently achieves at least 1.68x computational efficiency over current methods. By enabling energy-efficient CAM-based computing while preserving performance regardless of the array sparsity, MonoSparse-CAM addresses the high energy consumption problem of CAM which hinders processing of large arrays. Our contributions are twofold: we propose MonoSparse-CAM as an effective deployment optimization solution for CAM-based computing, and we investigate the impact of TBML model structure on array sparsity. This work provides crucial insights for energy-efficient TBML on hardware, highlighting a significant advancement in sustainable AI technologies.

Large language models (LLMs) have demonstrated remarkable proficiency in a range of natural language processing tasks. Once deployed, LLMs encounter users with personalized factual knowledge, and such personalized knowledge is consistently reflected through users' interactions with the LLMs. To enhance user experience, real-time model personalization is essential, allowing LLMs to adapt user-specific knowledge based on user feedback during human-LLM interactions. Existing methods mostly require back-propagation to finetune the model parameters, which incurs high computational and memory costs. In addition, these methods suffer from low interpretability, which will cause unforeseen impacts on model performance during long-term use, where the user's personalized knowledge is accumulated extensively.To address these challenges, we propose Knowledge Graph Tuning (KGT), a novel approach that leverages knowledge graphs (KGs) to personalize LLMs. KGT extracts personalized factual knowledge triples from users' queries and feedback and optimizes KGs without modifying the LLM parameters. Our method improves computational and memory efficiency by avoiding back-propagation and ensures interpretability by making the KG adjustments comprehensible to humans.Experiments with state-of-the-art LLMs, including GPT-2, Llama2, and Llama3, show that KGT significantly improves personalization performance while reducing latency and GPU memory costs. Ultimately, KGT offers a promising solution of effective, efficient, and interpretable real-time LLM personalization during user interactions with the LLMs.

Multiple visions of 6G networks elicit Artificial Intelligence (AI) as a central, native element. When 6G systems are deployed at a large scale, end-to-end AI-based solutions will necessarily have to encompass both the radio and the fiber-optical domain. This paper introduces the Decentralized Multi-Party, Multi-Network AI (DMMAI) framework for integrating AI into 6G networks deployed at scale. DMMAI harmonizes AI-driven controls across diverse network platforms and thus facilitates networks that autonomously configure, monitor, and repair themselves. This is particularly crucial at the network edge, where advanced applications meet heightened functionality and security demands. The radio/optical integration is vital due to the current compartmentalization of AI research within these domains, which lacks a comprehensive understanding of their interaction. Our approach explores multi-network orchestration and AI control integration, filling a critical gap in standardized frameworks for AI-driven coordination in 6G networks. The DMMAI framework is a step towards a global standard for AI in 6G, aiming to establish reference use cases, data and model management methods, and benchmarking platforms for future AI/ML solutions.

Forecasting the short-term spread of an ongoing disease outbreak is a formidable challenge due to the complexity of contributing factors, some of which can be characterized through interlinked, multi-modality variables such as epidemiological time series data, viral biology, population demographics, and the intersection of public policy and human behavior. Existing forecasting model frameworks struggle with the multifaceted nature of relevant data and robust results translation, which hinders their performances and the provision of actionable insights for public health decision-makers. Our work introduces PandemicLLM, a novel framework with multi-modal Large Language Models (LLMs) that reformulates real-time forecasting of disease spread as a text reasoning problem, with the ability to incorporate real-time, complex, non-numerical information that previously unattainable in traditional forecasting models. This approach, through a unique AI-human cooperative prompt design and time series representation learning, encodes multi-modal data for LLMs. The model is applied to the COVID-19 pandemic, and trained to utilize textual public health policies, genomic surveillance, spatial, and epidemiological time series data, and is subsequently tested across all 50 states of the U.S. Empirically, PandemicLLM is shown to be a high-performing pandemic forecasting framework that effectively captures the impact of emerging variants and can provide timely and accurate predictions. The proposed PandemicLLM opens avenues for incorporating various pandemic-related data in heterogeneous formats and exhibits performance benefits over existing models. This study illuminates the potential of adapting LLMs and representation learning to enhance pandemic forecasting, illustrating how AI innovations can strengthen pandemic responses and crisis management in the future., Comment: 35 pages, 10 figures

Dataset distillation (DD) has emerged as a widely adopted technique for crafting a synthetic dataset that captures the essential information of a training dataset, facilitating the training of accurate neural models. Its applications span various domains, including transfer learning, federated learning, and neural architecture search. The most popular methods for constructing the synthetic data rely on matching the convergence properties of training the model with the synthetic dataset and the training dataset. However, targeting the training dataset must be thought of as auxiliary in the same sense that the training set is an approximate substitute for the population distribution, and the latter is the data of interest. Yet despite its popularity, an aspect that remains unexplored is the relationship of DD to its generalization, particularly across uncommon subgroups. That is, how can we ensure that a model trained on the synthetic dataset performs well when faced with samples from regions with low population density? Here, the representativeness and coverage of the dataset become salient over the guaranteed training error at inference. Drawing inspiration from distributionally robust optimization, we introduce an algorithm that combines clustering with the minimization of a risk measure on the loss to conduct DD. We provide a theoretical rationale for our approach and demonstrate its effective generalization and robustness across subgroups through numerical experiments. The source code is available in https://github.com/Mming11/RobustDatasetDistillation.

Data heterogeneity presents significant challenges for federated learning (FL). Recently, dataset distillation techniques have been introduced, and performed at the client level, to attempt to mitigate some of these challenges. In this paper, we propose a highly efficient FL dataset distillation framework on the server side, significantly reducing both the computational and communication demands on local devices while enhancing the clients' privacy. Unlike previous strategies that perform dataset distillation on local devices and upload synthetic data to the server, our technique enables the server to leverage prior knowledge from pre-trained deep generative models to synthesize essential data representations from a heterogeneous model architecture. This process allows local devices to train smaller surrogate models while enabling the training of a larger global model on the server, effectively minimizing resource utilization. We substantiate our claim with a theoretical analysis, demonstrating the asymptotic resemblance of the process to the hypothetical ideal of completely centralized training on a heterogeneous dataset. Empirical evidence from our comprehensive experiments indicates our method's superiority, delivering an accuracy enhancement of up to 40% over non-dataset-distillation techniques in highly heterogeneous FL contexts, and surpassing existing dataset-distillation methods by 18%. In addition to the high accuracy, our framework converges faster than the baselines because rather than the server trains on several sets of heterogeneous data distributions, it trains on a multi-modal distribution. Our code is available at https://github.com/FedDG23/FedDG-main.git

Building on the cost-efficient pretraining advancements brought about by Crammed BERT, we enhance its performance and interpretability further by introducing a novel pretrained model Dependency Agreement Crammed BERT (DACBERT) and its two-stage pretraining framework - Dependency Agreement Pretraining. This framework, grounded by linguistic theories, seamlessly weaves syntax and semantic information into the pretraining process. The first stage employs four dedicated submodels to capture representative dependency agreements at the chunk level, effectively converting these agreements into embeddings. The second stage uses these refined embeddings, in tandem with conventional BERT embeddings, to guide the pretraining of the rest of the model. Evaluated on the GLUE benchmark, our DACBERT demonstrates notable improvement across various tasks, surpassing Crammed BERT by 3.13% in the RTE task and by 2.26% in the MRPC task. Furthermore, our method boosts the average GLUE score by 0.83%, underscoring its significant potential. The pretraining process can be efficiently executed on a single GPU within a 24-hour cycle, necessitating no supplementary computational resources or extending the pretraining duration compared with the Crammed BERT. Extensive studies further illuminate our approach's instrumental role in bolstering the interpretability of pretrained language models for natural language understanding tasks.

Pre-trained language models (PLM) have revolutionized the NLP landscape, achieving stellar performances across diverse tasks. These models, while benefiting from vast training data, often require fine-tuning on specific data to cater to distinct downstream tasks. However, this data adaptation process has inherent security and privacy concerns, primarily when leveraging user-generated, device-residing data. Federated learning (FL) provides a solution, allowing collaborative model fine-tuning without centralized data collection. However, applying FL to finetune PLMs is hampered by challenges, including restricted model parameter access, high computational requirements, and communication overheads. This paper introduces Federated Black-box Prompt Tuning (FedBPT), a framework designed to address these challenges. FedBPT does not require the clients to access the model parameters. By focusing on training optimal prompts and utilizing gradient-free optimization methods, FedBPT reduces the number of exchanged variables, boosts communication efficiency, and minimizes computational and storage costs. Experiments highlight the framework's ability to drastically cut communication and memory costs while maintaining competitive performance. Ultimately, FedBPT presents a promising solution for efficient, privacy-preserving fine-tuning of PLM in the age of large language models.

Deep neural networks (DNNs) have been successfully applied in various fields. A major challenge of deploying DNNs, especially on edge devices, is power consumption, due to the large number of multiply-and-accumulate (MAC) operations. To address this challenge, we propose PowerPruning, a novel method to reduce power consumption in digital neural network accelerators by selecting weights that lead to less power consumption in MAC operations. In addition, the timing characteristics of the selected weights together with all activation transitions are evaluated. The weights and activations that lead to small delays are further selected. Consequently, the maximum delay of the sensitized circuit paths in the MAC units is reduced even without modifying MAC units, which thus allows a flexible scaling of supply voltage to reduce power consumption further. Together with retraining, the proposed method can reduce power consumption of DNNs on hardware by up to 78.3% with only a slight accuracy loss.

Federated learning (FL) is a popular distributed learning framework that trains a global model through iterative communications between a central server and edge devices. Recent works have demonstrated that FL is vulnerable to model poisoning attacks. Several server-based defense approaches (e.g. robust aggregation), have been proposed to mitigate such attacks. However, we empirically show that under extremely strong attacks, these defensive methods fail to guarantee the robustness of FL. More importantly, we observe that as long as the global model is polluted, the impact of attacks on the global model will remain in subsequent rounds even if there are no subsequent attacks. In this work, we propose a client-based defense, named White Blood Cell for Federated Learning (FL-WBC), which can mitigate model poisoning attacks that have already polluted the global model. The key idea of FL-WBC is to identify the parameter space where long-lasting attack effect on parameters resides and perturb that space during local training. Furthermore, we derive a certified robustness guarantee against model poisoning attacks and a convergence guarantee to FedAvg after applying our FL-WBC. We conduct experiments on FasionMNIST and CIFAR10 to evaluate the defense against state-of-the-art model poisoning attacks. The results demonstrate that our method can effectively mitigate model poisoning attack impact on the global model within 5 communication rounds with nearly no accuracy drop under both IID and Non-IID settings. Our defense is also complementary to existing server-based robust aggregation approaches and can further improve the robustness of FL under extremely strong attacks., Comment: To be appeared in NeurIPS 2021 conference

Over recent years, a myriad of novel convolutional network architectures have been developed to advance state-of-the-art performance on challenging recognition tasks. As computational resources improve, a great deal of effort has been placed in efficiently scaling up existing designs and generating new architectures with Neural Architecture Search (NAS) algorithms. While network topology has proven to be a critical factor for model performance, we show that significant gains are being left on the table by keeping topology static at inference-time. Due to challenges such as scale variation, we should not expect static models configured to perform well across a training dataset to be optimally configured to handle all test data. In this work, we seek to expose the exciting potential of inference-time-dynamic models. By allowing just four layers to dynamically change configuration at inference-time, we show that existing off-the-shelf models like ResNet-50 are capable of over 95% accuracy on ImageNet. This level of performance currently exceeds that of models with over 20x more parameters and significantly more complex training procedures., Comment: 12 pages, 8 figures

Federated learning (FL) is a popular distributed learning framework that can reduce privacy risks by not explicitly sharing private data. However, recent works demonstrated that sharing model updates makes FL vulnerable to inference attacks. In this work, we show our key observation that the data representation leakage from gradients is the essential cause of privacy leakage in FL. We also provide an analysis of this observation to explain how the data presentation is leaked. Based on this observation, we propose a defense against model inversion attack in FL. The key idea of our defense is learning to perturb data representation such that the quality of the reconstructed data is severely degraded, while FL performance is maintained. In addition, we derive certified robustness guarantee to FL and convergence guarantee to FedAvg, after applying our defense. To evaluate our defense, we conduct experiments on MNIST and CIFAR10 for defending against the DLG attack and GS attack. Without sacrificing accuracy, the results demonstrate that our proposed defense can increase the mean squared error between the reconstructed data and the raw data by as much as more than 160X for both DLG attack and GS attack, compared with baseline defense methods. The privacy of the FL system is significantly improved.

Scale variance among different sizes of body parts and objects is a challenging problem for visual recognition tasks. Existing works usually design dedicated backbone or apply Neural architecture Search(NAS) for each task to tackle this challenge. However, existing works impose significant limitations on the design or search space. To solve these problems, we present ScaleNAS, a one-shot learning method for exploring scale-aware representations. ScaleNAS solves multiple tasks at a time by searching multi-scale feature aggregation. ScaleNAS adopts a flexible search space that allows an arbitrary number of blocks and cross-scale feature fusions. To cope with the high search cost incurred by the flexible space, ScaleNAS employs one-shot learning for multi-scale supernet driven by grouped sampling and evolutionary search. Without further retraining, ScaleNet can be directly deployed for different visual recognition tasks with superior performance. We use ScaleNAS to create high-resolution models for two different tasks, ScaleNet-P for human pose estimation and ScaleNet-S for semantic segmentation. ScaleNet-P and ScaleNet-S outperform existing manually crafted and NAS-based methods in both tasks. When applying ScaleNet-P to bottom-up human pose estimation, it surpasses the state-of-the-art HigherHRNet. In particular, ScaleNet-P4 achieves 71.6% AP on COCO test-dev, achieving new state-of-the-art result.

Link prediction in dynamic graphs (LPDG) is an important research problem that has diverse applications such as online recommendations, studies on disease contagion, organizational studies, etc. Various LPDG methods based on graph embedding and graph neural networks have been recently proposed and achieved state-of-the-art performance. In this paper, we study the vulnerability of LPDG methods and propose the first practical black-box evasion attack. Specifically, given a trained LPDG model, our attack aims to perturb the graph structure, without knowing to model parameters, model architecture, etc., such that the LPDG model makes as many wrong predicted links as possible. We design our attack based on a stochastic policy-based RL algorithm. Moreover, we evaluate our attack on three real-world graph datasets from different application domains. Experimental results show that our attack is both effective and efficient.

Neural Architecture Search (NAS) automates and prospers the design of neural networks. Estimator-based NAS has been proposed recently to model the relationship between architectures and their performance to enable scalable and flexible search. However, existing estimator-based methods encode the architecture into a latent space without considering graph similarity. Ignoring graph similarity in node-based search space may induce a large inconsistency between similar graphs and their distance in the continuous encoding space, leading to inaccurate encoding representation and/or reduced representation capacity that can yield sub-optimal search results. To preserve graph correlation information in encoding, we propose NASGEM which stands for Neural Architecture Search via Graph Embedding Method. NASGEM is driven by a novel graph embedding method equipped with similarity measures to capture the graph topology information. By precisely estimating the graph distance and using an auxiliary Weisfeiler-Lehman kernel to guide the encoding, NASGEM can utilize additional structural information to get more accurate graph representation to improve the search efficiency. GEMNet, a set of networks discovered by NASGEM, consistently outperforms networks crafted by existing search methods in classification tasks, i.e., with 0.4%-3.6% higher accuracy while having 11%- 21% fewer Multiply-Accumulates. We further transfer GEMNet for COCO object detection. In both one-stage and twostage detectors, our GEMNet surpasses its manually-crafted and automatically-searched counterparts.

Deep learning has been widely applied in many computer vision applications, with remarkable success. However, running deep learning models on mobile devices is generally challenging due to the limitation of computing resources. A popular alternative is to use cloud services to run deep learning models to process raw data. This, however, imposes privacy risks. Some prior arts proposed sending the features extracted from raw data to the cloud. Unfortunately, these extracted features can still be exploited by attackers to recover raw images and to infer embedded private attributes. In this paper, we propose an adversarial training framework, DeepObfuscator, which prevents the usage of the features for reconstruction of the raw images and inference of private attributes. This is done while retaining useful information for the intended cloud service. DeepObfuscator includes a learnable obfuscator that is designed to hide privacy-related sensitive information from the features by performing our proposed adversarial training algorithm. The proposed algorithm is designed by simulating the game between an attacker who makes efforts to reconstruct raw image and infer private attributes from the extracted features and a defender who aims to protect user privacy. By deploying the trained obfuscator on the smartphone, features can be locally extracted and then sent to the cloud. Our experiments on CelebA and LFW datasets show that the quality of the reconstructed images from the obfuscated features of the raw image is dramatically decreased from 0.9458 to 0.3175 in terms of multi-scale structural similarity. The person in the reconstructed image, hence, becomes hardly to be re-identified. The classification accuracy of the inferred private attributes that can be achieved by the attacker is significantly reduced to a random-guessing level., Comment: This paper is to be published in IoTDI'21

Designing neural architectures for edge devices is subject to constraints of accuracy, inference latency, and computational cost. Traditionally, researchers manually craft deep neural networks to meet the needs of mobile devices. Neural Architecture Search (NAS) was proposed to automate the neural architecture design without requiring extensive domain expertise and significant manual efforts. Recent works utilized NAS to design mobile models by taking into account hardware constraints and achieved state-of-the-art accuracy with fewer parameters and less computational cost measured in Multiply-accumulates (MACs). To find highly compact neural architectures, existing works relies on predefined cells and directly applying width multiplier, which may potentially limit the model flexibility, reduce the useful feature map information, and cause accuracy drop. To conquer this issue, we propose GRAM(GRAph propagation as Meta-knowledge) that adopts fine-grained (node-wise) search method and accumulates the knowledge learned in updates into a meta-graph. As a result, GRAM can enable more flexible search space and achieve higher search efficiency. Without the constraints of predefined cell or blocks, we propose a new structure-level pruning method to remove redundant operations in neural architectures. SwiftNet, which is a set of models discovered by GRAM, outperforms MobileNet-V2 by 2.15x higher accuracy density and 2.42x faster with similar accuracy. Compared with FBNet, SwiftNet reduces the search cost by 26x and achieves 2.35x higher accuracy density and 1.47x speedup while preserving similar accuracy. SwiftNetcan obtain 63.28% top-1 accuracy on ImageNet-1K with only 53M MACs and 2.07M parameters. The corresponding inference latency is only 19.09 ms on Google Pixel 1.

Computer vision has achieved impressive progress in recent years. Meanwhile, mobile phones have become the primary computing platforms for millions of people. In addition to mobile phones, many autonomous systems rely on visual data for making decisions and some of these systems have limited energy (such as unmanned aerial vehicles also called drones and mobile robots). These systems rely on batteries and energy efficiency is critical. This article serves two main purposes: (1) Examine the state-of-the-art for low-power solutions to detect objects in images. Since 2015, the IEEE Annual International Low-Power Image Recognition Challenge (LPIRC) has been held to identify the most energy-efficient computer vision solutions. This article summarizes 2018 winners' solutions. (2) Suggest directions for research as well as opportunities for low-power computer vision., Comment: Preprint, Accepted by IEEE Journal on Emerging and Selected Topics in Circuits and Systems. arXiv admin note: substantial text overlap with arXiv:1810.01732

Neural networks have shown great performance in cognitive tasks. When deploying network models on mobile devices with limited resources, weight quantization has been widely adopted. Binary quantization obtains the highest compression but usually results in big accuracy drop. In practice, 8-bit or 16-bit quantization is often used aiming at maintaining the same accuracy as the original 32-bit precision. We observe different layers have different accuracy sensitivity of quantization. Thus judiciously selecting different precision for different layers/structures can potentially produce more efficient models compared to traditional quantization methods by striking a better balance between accuracy and compression rate. In this work, we propose a fine-grained quantization approach for deep neural network compression by relaxing the search space of quantization bitwidth from discrete to a continuous domain. The proposed approach applies gradient descend based optimization to generate a mixed-precision quantization scheme that outperforms the accuracy of traditional quantization methods under the same compression rate., Comment: Hsin-Pai Cheng, Yuanjun Huang and Xuyang Guo contributed equally and are co-first authors for this paper. This work has been accepted by NIPS 2018 Workshop on Compact Deep Neural Network Representation with Industrial Applications, Montreal, Canada

Bayesian inference is an effective approach for solving statistical learning problems especially with uncertainty and incompleteness. However, inference efficiencies are physically limited by the bottlenecks of conventional computing platforms. In this paper, an emerging Bayesian inference system is proposed by exploiting spintronics based stochastic computing. A stochastic bitstream generator is realized as the kernel components by leveraging the inherent randomness of spintronics devices. The proposed system is evaluated by typical applications of data fusion and Bayesian belief networks. Simulation results indicate that the proposed approach could achieve significant improvement on inference efficiencies in terms of power consumption and inference speed., Comment: accepted by ASPDAC 2018 conference

Model compression is significant for the wide adoption of Recurrent Neural Networks (RNNs) in both user devices possessing limited resources and business clusters requiring quick responses to large-scale service requests. This work aims to learn structurally-sparse Long Short-Term Memory (LSTM) by reducing the sizes of basic structures within LSTM units, including input updates, gates, hidden states, cell states and outputs. Independently reducing the sizes of basic structures can result in inconsistent dimensions among them, and consequently, end up with invalid LSTM units. To overcome the problem, we propose Intrinsic Sparse Structures (ISS) in LSTMs. Removing a component of ISS will simultaneously decrease the sizes of all basic structures by one and thereby always maintain the dimension consistency. By learning ISS within LSTM units, the obtained LSTMs remain regular while having much smaller basic structures. Based on group Lasso regularization, our method achieves 10.59x speedup without losing any perplexity of a language modeling of Penn TreeBank dataset. It is also successfully evaluated through a compact model with only 2.69M weights for machine Question Answering of SQuAD dataset. Our approach is successfully extended to non- LSTM RNNs, like Recurrent Highway Networks (RHNs). Our source code is publicly available at https://github.com/wenwei202/iss-rnns, Comment: Published in ICLR 2018 ( the Sixth International Conference on Learning Representations)

Recently, DNN model compression based on network architecture design, e.g., SqueezeNet, attracted a lot attention. No accuracy drop on image classification is observed on these extremely compact networks, compared to well-known models. An emerging question, however, is whether these model compression techniques hurt DNN's learning ability other than classifying images on a single dataset. Our preliminary experiment shows that these compression methods could degrade domain adaptation (DA) ability, though the classification performance is preserved. Therefore, we propose a new compact network architecture and unsupervised DA method in this paper. The DNN is built on a new basic module Conv-M which provides more diverse feature extractors without significantly increasing parameters. The unified framework of our DA method will simultaneously learn invariance across domains, reduce divergence of feature representations, and adapt label prediction. Our DNN has 4.1M parameters, which is only 6.7% of AlexNet or 59% of GoogLeNet. Experiments show that our DNN obtains GoogLeNet-level accuracy both on classification and DA, and our DA method slightly outperforms previous competitive ones. Put all together, our DA strategy based on our DNN achieves state-of-the-art on sixteen of total eighteen DA tasks on popular Office-31 and Office-Caltech datasets., Comment: 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR'17)

IBM TrueNorth chip uses digital spikes to perform neuromorphic computing and achieves ultrahigh execution parallelism and power efficiency. However, in TrueNorth chip, low quantization resolution of the synaptic weights and spikes significantly limits the inference (e.g., classification) accuracy of the deployed neural network model. Existing workaround, i.e., averaging the results over multiple copies instantiated in spatial and temporal domains, rapidly exhausts the hardware resources and slows down the computation. In this work, we propose a novel learning method on TrueNorth platform that constrains the random variance of each computation copy and reduces the number of needed copies. Compared to the existing learning method, our method can achieve up to 68.8% reduction of the required neuro-synaptic cores or 6.5X speedup, with even slightly improved inference accuracy., Comment: 6 pages; 9 figures; DAC 2016