SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MB model size

Abstract

Recent research on deep neural networks has focused primarily on improving accuracy. For a given accuracy level, it is typically possible to identify multiple DNN architectures that achieve that accuracy level. With equivalent accuracy, smaller DNN architectures offer at least three advantages: (1) Smaller DNNs require less communication across servers during distributed training. (2) Smaller DNNs require less bandwidth to export a new model from the cloud to an autonomous car. (3) Smaller DNNs are more feasible to deploy on FPGAs and other hardware with limited memory. To provide all of these advantages, we propose a small DNN architecture called SqueezeNet. SqueezeNet achieves AlexNet-level accuracy on ImageNet with 50x fewer parameters. Additionally, with model compression techniques we are able to compress SqueezeNet to less than 0.5MB (510x smaller than AlexNet). The SqueezeNet architecture is available for download here: https://github.com/DeepScale/SqueezeNet

In ICLR Format

Cited by in corpus (418)

• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
• A Survey of Deep Learning Techniques for Autonomous Driving
• Convergence of Edge Computing and Deep Learning: A Comprehensive Survey
• Explaining Deep Neural Networks and Beyond: A Review of Methods and Applications
• Deep Multi-modal Object Detection and Semantic Segmentation for Autonomous Driving: Datasets, Methods, and Challenges
• A Survey of Deep Learning-based Object Detection
• The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks
• Deep Face Recognition: A Survey
• Benchmark Analysis of Representative Deep Neural Network Architectures
• The Unreasonable Effectiveness of Deep Features as a Perceptual Metric
• Ternary Weight Networks
• Deep Learning in Video Multi-Object Tracking: A Survey
• FractalNet: Ultra-Deep Neural Networks without Residuals
• Object Detection in 20 Years: A Survey
• The History Began from AlexNet: A Comprehensive Survey on Deep Learning Approaches
• Visual Transformers: Token-based Image Representation and Processing for Computer Vision
• Bayesian Compression for Deep Learning
• Point-Voxel CNN for Efficient 3D Deep Learning
• A Light CNN for detecting COVID-19 from CT scans of the chest
• MixConv: Mixed Depthwise Convolutional Kernels
• Deep Roots: Improving CNN Efficiency with Hierarchical Filter Groups
• Machine Learning for Wireless Communications in the Internet of Things: A Comprehensive Survey
• ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware
• XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks
• Hardware-oriented Approximation of Convolutional Neural Networks
• Model compression via distillation and quantization
• Deep Learning for Generic Object Detection: A Survey
• Benchmarking TPU, GPU, and CPU Platforms for Deep Learning
• Strengths and Weaknesses of Deep Learning Models for Face Recognition Against Image Degradations
• Demystifying Parallel and Distributed Deep Learning: An In-Depth Concurrency Analysis
• Quantization and Deployment of Deep Neural Networks on Microcontrollers
• Group Knowledge Transfer: Federated Learning of Large CNNs at the Edge
• Dynamic Convolutions: Exploiting Spatial Sparsity for Faster Inference
• Towards the Systematic Reporting of the Energy and Carbon Footprints of Machine Learning
• TSM: Temporal Shift Module for Efficient Video Understanding
• CornerNet-Lite: Efficient Keypoint Based Object Detection
• Deep Learning in the Automotive Industry: Applications and Tools
• Stabilizing the Lottery Ticket Hypothesis
• Layer-wise training convolutional neural networks with smaller filters for human activity recognition using wearable sensors
• A Survey of FPGA-Based Neural Network Accelerator
• Enable Deep Learning on Mobile Devices: Methods, Systems, and Applications
• Computer Vision for Road Imaging and Pothole Detection: A State-of-the-Art Review of Systems and Algorithms
• High-Throughput CNN Inference on Embedded ARM big.LITTLE Multi-Core Processors
• XNOR Neural Engine: a Hardware Accelerator IP for 21.6 fJ/op Binary Neural Network Inference
• Joint 3D Proposal Generation and Object Detection from View Aggregation
• FastDeepIoT: Towards Understanding and Optimizing Neural Network Execution Time on Mobile and Embedded Devices
• Deep Learning in Mobile and Wireless Networking: A Survey
• ICNet for Real-Time Semantic Segmentation on High-Resolution Images
• Edge Intelligence: Architectures, Challenges, and Applications
• BiSeNet: Bilateral Segmentation Network for Real-time Semantic Segmentation
• Compressing Neural Networks using the Variational Information Bottleneck
• An Accurate EEGNet-based Motor-Imagery Brain-Computer Interface for Low-Power Edge Computing
• Omni-Scale Feature Learning for Person Re-Identification
• Exploring the Connection Between Binary and Spiking Neural Networks
• GhostNet: More Features from Cheap Operations
• Recent Advances in Deep Learning: An Overview
• Machine learning and AI-based approaches for bioactive ligand discovery and GPCR-ligand recognition
• Recent Advances in Object Detection in the Age of Deep Convolutional Neural Networks
• LPRNet: License Plate Recognition via Deep Neural Networks
• Light-Weight RefineNet for Real-Time Semantic Segmentation
• A Survey on Approximate Edge AI for Energy Efficient Autonomous Driving Services
• Hardware Acceleration of Sparse and Irregular Tensor Computations of ML Models: A Survey and Insights
• Training Convolutional Neural Networks with Limited Training Data for Ear Recognition in the Wild
• Siamese Attentional Keypoint Network for High Performance Visual Tracking
• PointSeg: Real-Time Semantic Segmentation Based on 3D LiDAR Point Cloud
• Decoding and mapping task states of the human brain via deep learning
• Tiny-DSOD: Lightweight Object Detection for Resource-Restricted Usages
• Ristretto: Hardware-Oriented Approximation of Convolutional Neural Networks
• Once-for-All: Train One Network and Specialize it for Efficient Deployment
• MetaPruning: Meta Learning for Automatic Neural Network Channel Pruning
• CondenseNet: An Efficient DenseNet using Learned Group Convolutions
• Accelerator-Aware Pruning for Convolutional Neural Networks
• Deep $k$-Means: Re-Training and Parameter Sharing with Harder Cluster Assignments for Compressing Deep Convolutions
• Loss-aware Weight Quantization of Deep Networks
• Selection via Proxy: Efficient Data Selection for Deep Learning
• Dual Dynamic Inference: Enabling More Efficient, Adaptive and Controllable Deep Inference
• FBNet: Hardware-Aware Efficient ConvNet Design via Differentiable Neural Architecture Search
• ChineseFoodNet: A large-scale Image Dataset for Chinese Food Recognition
• What Do Compressed Deep Neural Networks Forget?
• Paraphrasing Complex Network: Network Compression via Factor Transfer
• A Comprehensive Overview on UWB Radar: Applications, Standards, Signal Processing Techniques, Datasets, Radio Chips, Trends and Future Research Directions
• SkyNet: a Hardware-Efficient Method for Object Detection and Tracking on Embedded Systems
• Evolving Deep Convolutional Neural Networks for Image Classification
• Testing Robustness Against Unforeseen Adversaries
• A Low Effort Approach to Structured CNN Design Using PCA
• Face Deidentification with Generative Deep Neural Networks
• Pruning neural networks without any data by iteratively conserving synaptic flow
• MeliusNet: Can Binary Neural Networks Achieve MobileNet-level Accuracy?
• Correction of AI systems by linear discriminants: Probabilistic foundations
• Interpretation of Neural Networks is Fragile
• Dynamic Convolution: Attention over Convolution Kernels
• Domain Adaptation for Ear Recognition Using Deep Convolutional Neural Networks
• A novel channel pruning method for deep neural network compression
• The unreasonable effectiveness of small neural ensembles in high-dimensional brain
• Efficient Multi-objective Neural Architecture Search via Lamarckian Evolution
• ECOVNet: An Ensemble of Deep Convolutional Neural Networks Based on EfficientNet to Detect COVID-19 From Chest X-rays
• And the Bit Goes Down: Revisiting the Quantization of Neural Networks
• Neuron Shapley: Discovering the Responsible Neurons
• RMPE: Regional Multi-person Pose Estimation
• EDropout: Energy-Based Dropout and Pruning of Deep Neural Networks
• Measuring the Intrinsic Dimension of Objective Landscapes
• Acoustic Anomaly Detection for Machine Sounds based on Image Transfer Learning
• Gaussian Splatting: 3D Reconstruction and Novel View Synthesis, a Review
• Structured Probabilistic Pruning for Convolutional Neural Network Acceleration
• FermiNets: Learning generative machines to generate efficient neural networks via generative synthesis
• Lets keep it simple, Using simple architectures to outperform deeper and more complex architectures
• WiFiNet: WiFi-based indoor localisation using CNNs
• Q-BERT: Hessian Based Ultra Low Precision Quantization of BERT
• SS-CAM: Smoothed Score-CAM for Sharper Visual Feature Localization
• Edge Intelligence: On-Demand Deep Learning Model Co-Inference with Device-Edge Synergy
• Video Semantic Segmentation with Distortion-Aware Feature Correction
• A Survey on IoT Ground Sensing Systems for Early Wildfire Detection: Technologies, Challenges and Opportunities
• Learning Generalisable Omni-Scale Representations for Person Re-Identification
• Deep Learning for 3D Point Clouds: A Survey
• Winning the Lottery with Continuous Sparsification
• On-Device Machine Learning: An Algorithms and Learning Theory Perspective
• MobileFaceNets: Efficient CNNs for Accurate Real-Time Face Verification on Mobile Devices
• Constructing Energy-efficient Mixed-precision Neural Networks through Principal Component Analysis for Edge Intelligence
• Efficient Processing of Deep Neural Networks: A Tutorial and Survey
• Diversity can be Transferred: Output Diversification for White- and Black-box Attacks
• Unite the People: Closing the Loop Between 3D and 2D Human Representations
• An Empirical Study on Writer Identification & Verification from Intra-variable Individual Handwriting
• Mobile Video Object Detection with Temporally-Aware Feature Maps
• YOLO-LITE: A Real-Time Object Detection Algorithm Optimized for Non-GPU Computers
• BlockDrop: Dynamic Inference Paths in Residual Networks
• Knowledge Transfer Between Artificial Intelligence Systems
• More is Less: A More Complicated Network with Less Inference Complexity
• DCFNet: Deep Neural Network with Decomposed Convolutional Filters
• Low-Memory Neural Network Training: A Technical Report
• Towards High Performance Video Object Detection for Mobiles
• HAWQ: Hessian AWare Quantization of Neural Networks with Mixed-Precision
• Pixel-wise Attentional Gating for Parsimonious Pixel Labeling
• Deep Layer Aggregation
• TinyTL: Reduce Activations, Not Trainable Parameters for Efficient On-Device Learning
• An embedded system for the automated generation of labeled plant images to enable machine learning applications in agriculture
• Training Competitive Binary Neural Networks from Scratch
• Fixed-Point Convolutional Neural Network for Real-Time Video Processing in FPGA
• FINN-R: An End-to-End Deep-Learning Framework for Fast Exploration of Quantized Neural Networks
• DeProCams: Simultaneous Relighting, Compensation and Shape Reconstruction for Projector-Camera Systems
• Musical Chair: Efficient Real-Time Recognition Using Collaborative IoT Devices
• Combating the Elsagate phenomenon: Deep learning architectures for disturbing cartoons
• Data-Driven Sparse Structure Selection for Deep Neural Networks
• SqueezeSegV3: Spatially-Adaptive Convolution for Efficient Point-Cloud Segmentation
• RefineDetLite: A Lightweight One-stage Object Detection Framework for CPU-only Devices
• RNNPool: Efficient Non-linear Pooling for RAM Constrained Inference
• IGCV3: Interleaved Low-Rank Group Convolutions for Efficient Deep Neural Networks
• Privado: Practical and Secure DNN Inference with Enclaves
• Survey on the Analysis and Modeling of Visual Kinship: A Decade in the Making
• Material Recognition for Automated Progress Monitoring using Deep Learning Methods
• IamNN: Iterative and Adaptive Mobile Neural Network for Efficient Image Classification
• Computer-aided diagnosis of lung carcinoma using deep learning - a pilot study
• Nimble: Efficiently Compiling Dynamic Neural Networks for Model Inference
• Task dependent Deep LDA pruning of neural networks
• Disentangling Image Distortions in Deep Feature Space
• COVID-MobileXpert: On-Device COVID-19 Patient Triage and Follow-up using Chest X-rays
• Quantization for Rapid Deployment of Deep Neural Networks
• VarGNet: Variable Group Convolutional Neural Network for Efficient Embedded Computing
• Constructing Fast Network through Deconstruction of Convolution
• Families In Wild Multimedia: A Multimodal Database for Recognizing Kinship
• Bi-Real Net: Enhancing the Performance of 1-bit CNNs With Improved Representational Capability and Advanced Training Algorithm
• Assigning a Grade: Accurate Measurement of Road Quality Using Satellite Imagery
• Towards Effective Low-bitwidth Convolutional Neural Networks
• NeuPart: Using Analytical Models to Drive Energy-Efficient Partitioning of CNN Computations on Cloud-Connected Mobile Clients
• A Novel Hybrid Machine Learning Model for Auto-Classification of Retinal Diseases
• Model Slicing for Supporting Complex Analytics with Elastic Inference Cost and Resource Constraints
• Lightweight Residual Densely Connected Convolutional Neural Network
• Resource-Efficient Neural Networks for Embedded Systems
• Structural Compression of Convolutional Neural Networks
• PaveSAM Segment Anything for Pavement Distress
• Characterizing the Deep Neural Networks Inference Performance of Mobile Applications
• AutoQ: Automated Kernel-Wise Neural Network Quantization
• Quantizing Convolutional Neural Networks for Low-Power High-Throughput Inference Engines
• A Less Biased Evaluation of Out-of-distribution Sample Detectors
• Sparse Transfer Learning via Winning Lottery Tickets
• Rate Distortion For Model Compression: From Theory To Practice
• Joint Neural Architecture Search and Quantization
• ZeroQ: A Novel Zero Shot Quantization Framework
• DeepHammer: Depleting the Intelligence of Deep Neural Networks through Targeted Chain of Bit Flips
• Bit-Tactical: Exploiting Ineffectual Computations in Convolutional Neural Networks: Which, Why, and How
• Finding Better Topologies for Deep Convolutional Neural Networks by Evolution
• Fast On-the-fly Retraining-free Sparsification of Convolutional Neural Networks
• A model-driven approach for a new generation of adaptive libraries
• Fast Construction of Correcting Ensembles for Legacy Artificial Intelligence Systems: Algorithms and a Case Study
• Measuring and regularizing networks in function space
• A Loss Function for Generative Neural Networks Based on Watson's Perceptual Model
• Fairness Through Robustness: Investigating Robustness Disparity in Deep Learning
• Structured Binary Neural Networks for Accurate Image Classification and Semantic Segmentation
• Percival: Making In-Browser Perceptual Ad Blocking Practical With Deep Learning
• Multimodal Detection of Bots on X (Twitter) using Transformers
• Sparse Deep Neural Network Exact Solutions
• Scratch that! An Evolution-based Adversarial Attack against Neural Networks
• Learning to Train a Binary Neural Network
• Structured Knowledge Distillation for Dense Prediction
• American Sign Language Alphabet Recognition using Deep Learning
• Extract-Transform-Load for Video Streams
• SiPPing Neural Networks: Sensitivity-informed Provable Pruning of Neural Networks
• ExSample: Efficient Searches on Video Repositories through Adaptive Sampling
• Rafiki: Machine Learning as an Analytics Service System
• Recognizing Birds from Sound - The 2018 BirdCLEF Baseline System
• MoBiNet: A Mobile Binary Network for Image Classification
• Binary Ensemble Neural Network: More Bits per Network or More Networks per Bit?
• VSGNet: Spatial Attention Network for Detecting Human Object Interactions Using Graph Convolutions
• Re-Identification with Consistent Attentive Siamese Networks
• Exploiting Errors for Efficiency: A Survey from Circuits to Algorithms
• Mobile Video Action Recognition
• EffNet: An Efficient Structure for Convolutional Neural Networks
• Rethinking Channel Dimensions for Efficient Model Design
• AdderNet: Do We Really Need Multiplications in Deep Learning?
• Automated detection of oral pre-cancerous tongue lesions using deep learning for early diagnosis of oral cavity cancer
• Searching for Winograd-aware Quantized Networks
• FD-MobileNet: Improved MobileNet with a Fast Downsampling Strategy
• Learning Anytime Predictions in Neural Networks via Adaptive Loss Balancing
• A Study of Deep Learning Robustness Against Computation Failures
• Pack and Detect: Fast Object Detection in Videos Using Region-of-Interest Packing
• Study and development of a Computer-Aided Diagnosis system for classification of chest x-ray images using convolutional neural networks pre-trained for ImageNet and data augmentation
• Shift-based Primitives for Efficient Convolutional Neural Networks
• Road Segmentation Using CNN with GRU
• RotDCF: Decomposition of Convolutional Filters for Rotation-Equivariant Deep Networks
• LCNN: Lookup-based Convolutional Neural Network
• Fused DNN: A deep neural network fusion approach to fast and robust pedestrian detection
• DAC-SDC Low Power Object Detection Challenge for UAV Applications
• A Framework of Transfer Learning in Object Detection for Embedded Systems
• Optimizing Video Object Detection via a Scale-Time Lattice
• BWCNN: Blink to Word, a Real-Time Convolutional Neural Network Approach
• GraphChallenge.org Sparse Deep Neural Network Performance
• Optimization of FPGA-based CNN Accelerators Using Metaheuristics
• Eyeriss v2: A Flexible Accelerator for Emerging Deep Neural Networks on Mobile Devices
• Fast Video Shot Transition Localization with Deep Structured Models
• Automated Design Space Exploration for optimised Deployment of DNN on Arm Cortex-A CPUs
• Land Classification in Satellite Images by Injecting Traditional Features to CNN Models
• Ensemble Knowledge Distillation for Learning Improved and Efficient Networks
• Efficient Deep Neural Networks
• AI Blue Book: Vehicle Price Prediction using Visual Features
• ApproxNet: Content and Contention-Aware Video Analytics System for Embedded Clients
• Learning Deep Representations with Probabilistic Knowledge Transfer
• Organ-based Chronological Age Estimation based on 3D MRI Scans
• Effective Training of Convolutional Neural Networks with Low-bitwidth Weights and Activations
• Compact Global Descriptor for Neural Networks
• TKD: Temporal Knowledge Distillation for Active Perception
• CHET: Compiler and Runtime for Homomorphic Evaluation of Tensor Programs
• Training Compact Neural Networks with Binary Weights and Low Precision Activations
• Deep Neural Networks with Multi-Branch Architectures Are Less Non-Convex
• Rethinking Machine Learning Development and Deployment for Edge Devices
• Espresso: Efficient Forward Propagation for BCNNs
• Growing Efficient Deep Networks by Structured Continuous Sparsification
• Oriented Boxes for Accurate Instance Segmentation
• Boosting Black-Box Attack with Partially Transferred Conditional Adversarial Distribution
• Benanza: Automatic $\mu$Benchmark Generation to Compute "Lower-bound" Latency and Inform Optimizations of Deep Learning Models on GPUs
• Bi-Real Net: Binarizing Deep Network Towards Real-Network Performance
• Exploring the Intersection between Neural Architecture Search and Continual Learning
• Remote Sensor Design for Visual Recognition with Convolutional Neural Networks
• FDFtNet: Facing Off Fake Images using Fake Detection Fine-tuning Network
• Residual Squeeze VGG16
• Co-Design of Deep Neural Nets and Neural Net Accelerators for Embedded Vision Applications
• HSD-CNN: Hierarchically self decomposing CNN architecture using class specific filter sensitivity analysis
• Factorized Distillation: Training Holistic Person Re-identification Model by Distilling an Ensemble of Partial ReID Models
• Fast Object Detection in Compressed Video
• An Evaluation of Deep CNN Baselines for Scene-Independent Person Re-Identification
• Learning to Zoom: a Saliency-Based Sampling Layer for Neural Networks
• Feature Selective Small Object Detection via Knowledge-based Recurrent Attentive Neural Network
• PANDA: Facilitating Usable AI Development
• BrainSlug: Transparent Acceleration of Deep Learning Through Depth-First Parallelism
• Collaborative Distillation for Ultra-Resolution Universal Style Transfer
• Generative Model with Coordinate Metric Learning for Object Recognition Based on 3D Models
• RILOD: Near Real-Time Incremental Learning for Object Detection at the Edge
• Detecting The Objects on The Road Using Modular Lightweight Network
• Learning From Less Data: Diversified Subset Selection and Active Learning in Image Classification Tasks
• Enabling Embedded Inference Engine with ARM Compute Library: A Case Study
• Hazard Detection in Supermarkets using Deep Learning on the Edge
• GXNOR-Net: Training deep neural networks with ternary weights and activations without full-precision memory under a unified discretization framework
• Increasing Shape Bias in ImageNet-Trained Networks Using Transfer Learning and Domain-Adversarial Methods
• Data augmentation with Symbolic-to-Real Image Translation GANs for Traffic Sign Recognition
• HarDNet: A Low Memory Traffic Network
• Seesaw-Net: Convolution Neural Network With Uneven Group Convolution
• Tag Prediction at Flickr: a View from the Darkroom
• Learning Strict Identity Mappings in Deep Residual Networks
• KS(conf ): A Light-Weight Test if a ConvNet Operates Outside of Its Specifications
• Interactive Classification for Deep Learning Interpretation
• Discriminator Feature-based Inference by Recycling the Discriminator of GANs
• How deep should be the depth of convolutional neural networks: a backyard dog case study
• Spectral Pruning: Compressing Deep Neural Networks via Spectral Analysis and its Generalization Error
• Trace norm regularization and faster inference for embedded speech recognition RNNs
• Data-Dependent Coresets for Compressing Neural Networks with Applications to Generalization Bounds
• Fire SSD: Wide Fire Modules based Single Shot Detector on Edge Device
• Dynamic Runtime Feature Map Pruning
• Learning Architectures for Binary Networks
• Harmonic Adversarial Attack Method
• Multigrid Neural Architectures
• Evolving SimGANs to Improve Abnormal Electrocardiogram Classification
• QuTiBench: Benchmarking Neural Networks on Heterogeneous Hardware
• Testing Deep Learning Models for Image Analysis Using Object-Relevant Metamorphic Relations
• Learning towards Minimum Hyperspherical Energy
• The Utility of Feature Reuse: Transfer Learning in Data-Starved Regimes
• The Effect of Learning Strategy versus Inherent Architecture Properties on the Ability of Convolutional Neural Networks to Develop Transformation Invariance
• Adaptive Selection of Deep Learning Models on Embedded Systems
• Sparsely Aggregated Convolutional Networks
• DPRed: Making Typical Activation and Weight Values Matter In Deep Learning Computing
• You Only Spike Once: Improving Energy-Efficient Neuromorphic Inference to ANN-Level Accuracy
• LCP: A Low-Communication Parallelization Method for Fast Neural Network Inference in Image Recognition
• Convolutional Neural Network Quantization using Generalized Gamma Distribution
• A Benchmark for Temporal Color Constancy
• Neural Network Compression using Transform Coding and Clustering
• PydMobileNet: Improved Version of MobileNets with Pyramid Depthwise Separable Convolution
• A Survey of Mobile Computing for the Visually Impaired
• Breast Cancer Classification from Histopathological Images with Inception Recurrent Residual Convolutional Neural Network
• Are We Hungry for 3D LiDAR Data for Semantic Segmentation? A Survey and Experimental Study
• Domain Adaptation with Joint Learning for Generic, Optical Car Part Recognition and Detection Systems (Go-CaRD)
• VACL: Variance-Aware Cross-Layer Regularization for Pruning Deep Residual Networks
• MosAIc: Finding Artistic Connections across Culture with Conditional Image Retrieval
• Zero-Shot Deep Domain Adaptation
• SeesawFaceNets: sparse and robust face verification model for mobile platform
• Ultrafast-and-Ultralight ConvNet-Based Intelligent Monitoring System for Diagnosing Early-Stage Mpox Anytime and Anywhere
• An FPGA-Accelerated Design for Deep Learning Pedestrian Detection in Self-Driving Vehicles
• Graph-Adaptive Pruning for Efficient Inference of Convolutional Neural Networks
• Impostor Networks for Fast Fine-Grained Recognition
• PointAR: Efficient Lighting Estimation for Mobile Augmented Reality
• On a Sparse Shortcut Topology of Artificial Neural Networks
• Digital Neuron: A Hardware Inference Accelerator for Convolutional Deep Neural Networks
• ReXCam: Resource-Efficient, Cross-Camera Video Analytics at Scale
• 3D latent diffusion models for parameterizing and history matching multiscenario facies systems
• A First Look at Deep Learning Apps on Smartphones
• Merging and Evolution: Improving Convolutional Neural Networks for Mobile Applications
• SlideNet: Fast and Accurate Slide Quality Assessment Based on Deep Neural Networks
• Joint Architecture and Knowledge Distillation in CNN for Chinese Text Recognition
• Energy-Aware DNN Graph Optimization
• Optimizing Neural Architecture Search using Limited GPU Time in a Dynamic Search Space: A Gene Expression Programming Approach
• Activation Density driven Energy-Efficient Pruning in Training
• FrostNet: Towards Quantization-Aware Network Architecture Search
• Recent Advances in Efficient Computation of Deep Convolutional Neural Networks
• Influence of Image Classification Accuracy on Saliency Map Estimation
• Performance Analysis and Characterization of Training Deep Learning Models on Mobile Devices
• Sketch-Specific Data Augmentation for Freehand Sketch Recognition
• Analyzing Machine Learning Workloads Using a Detailed GPU Simulator
• MemNet: Memory-Efficiency Guided Neural Architecture Search with Augment-Trim learning
• Cloud-based or On-device: An Empirical Study of Mobile Deep Inference
• Fabrik: An Online Collaborative Neural Network Editor
• A Convolution Tree with Deconvolution Branches: Exploiting Geometric Relationships for Single Shot Keypoint Detection
• A Compact DNN: Approaching GoogLeNet-Level Accuracy of Classification and Domain Adaptation
• Image and Texture Independent Deep Learning Noise Estimation using Multiple Frames
• Refining the Structure of Neural Networks Using Matrix Conditioning
• MDCN: Multi-Scale, Deep Inception Convolutional Neural Networks for Efficient Object Detection
• SECS: Efficient Deep Stream Processing via Class Skew Dichotomy
• Toward Autonomous Rotation-Aware Unmanned Aerial Grasping
• Maximizing CNN Accelerator Efficiency Through Resource Partitioning
• Tuning Algorithms and Generators for Efficient Edge Inference
• Fine-Grained Energy and Performance Profiling framework for Deep Convolutional Neural Networks
• Object Detection with Mask-based Feature Encoding
• Supervised Deep Sparse Coding Networks
• PointHop++: A Lightweight Learning Model on Point Sets for 3D Classification
• Derivation and Analysis of Fast Bilinear Algorithms for Convolution
• Stealing the Invisible: Unveiling Pre-Trained CNN Models through Adversarial Examples and Timing Side-Channels
• ModiPick: SLA-aware Accuracy Optimization For Mobile Deep Inference
• Kerman: A Hybrid Lightweight Tracking Algorithm to Enable Smart Surveillance as an Edge Service
• DNN Feature Map Compression using Learned Representation over GF(2)
• Sparsity in Deep Neural Networks - An Empirical Investigation with TensorQuant
• Deep Global-Connected Net With The Generalized Multi-Piecewise ReLU Activation in Deep Learning
• HBONet: Harmonious Bottleneck on Two Orthogonal Dimensions
• Dense xUnit Networks
• WaveletNet: Logarithmic Scale Efficient Convolutional Neural Networks for Edge Devices
• Data Analytics Service Composition and Deployment on Edge Devices
• MDInference: Balancing Inference Accuracy and Latency for Mobile Applications
• Diagonalwise Refactorization: An Efficient Training Method for Depthwise Convolutions
• Transfer Incremental Learning using Data Augmentation
• Local Binary Pattern Networks
• Reaching Data Confidentiality and Model Accountability on the CalTrain
• Real-time Burst Photo Selection Using a Light-Head Adversarial Network
• MPDCompress - Matrix Permutation Decomposition Algorithm for Deep Neural Network Compression
• ISP4ML: Understanding the Role of Image Signal Processing in Efficient Deep Learning Vision Systems
• Robust Student Network Learning
• Compact Deep Neural Networks for Computationally Efficient Gesture Classification From Electromyography Signals
• PSDNet and DPDNet: Efficient channel expansion, Depthwise-Pointwise-Depthwise Inverted Bottleneck Block
• Context-aware Cascade Attention-based RNN for Video Emotion Recognition
• RRNet: Repetition-Reduction Network for Energy Efficient Decoder of Depth Estimation
• Affine Multiplexing Networks: System Analysis, Learning, and Computation
• Analyzing Visual Representations in Embodied Navigation Tasks
• BEAN: Interpretable Representation Learning with Biologically-Enhanced Artificial Neuronal Assembly Regularization
• Compression and Acceleration of Neural Networks for Communications
• Machine Learning Systems for Intelligent Services in the IoT: A Survey
• Improved Semantic Stixels via Multimodal Sensor Fusion
• Efficient and Robust Machine Learning for Real-World Systems
• Spectral Image Visualization Using Generative Adversarial Networks
• PBGen: Partial Binarization of Deconvolution-Based Generators for Edge Intelligence
• A Miniaturized Semantic Segmentation Method for Remote Sensing Image
• A mixed signal architecture for convolutional neural networks
• Dynamic Routing Networks
• Crack-pot: Autonomous Road Crack and Pothole Detection
• The Unreasonable Effectiveness of Texture Transfer for Single Image Super-resolution
• ExGate: Externally Controlled Gating for Feature-based Attention in Artificial Neural Networks
• ShuffleDet: Real-Time Vehicle Detection Network in On-board Embedded UAV Imagery
• Simplifying Neural Networks using Formal Verification
• Learning Blind Video Temporal Consistency
• Tuned Compositional Feature Replays for Efficient Stream Learning
• Adjoined Networks: A Training Paradigm with Applications to Network Compression
• Auto-Classification of Retinal Diseases in the Limit of Sparse Data Using a Two-Streams Machine Learning Model
• Fast Retinomorphic Event Stream for Video Recognition and Reinforcement Learning
• Group Pruning using a Bounded-Lp norm for Group Gating and Regularization
• Performance tuning for deep learning on a many-core processor (master thesis)
• LO-Net: Deep Real-time Lidar Odometry
• Light Multi-segment Activation for Model Compression
• Deep Learning Approximation: Zero-Shot Neural Network Speedup
• Attention-Guided Lightweight Network for Real-Time Segmentation of Robotic Surgical Instruments
• BRIEF: Backward Reduction of CNNs with Information Flow Analysis
• Hoplite: Efficient and Fault-Tolerant Collective Communication for Task-Based Distributed Systems
• Training Domain Specific Models for Energy-Efficient Object Detection
• Elastic Neural Networks: A Scalable Framework for Embedded Computer Vision
• Learning to Prevent Leakage: Privacy-Preserving Inference in the Mobile Cloud
• What Happens on the Edge, Stays on the Edge: Toward Compressive Deep Learning
• Classification of Chinese Handwritten Numbers with Labeled Projective Dictionary Pair Learning
• Pacemaker: Intermediate Teacher Knowledge Distillation For On-The-Fly Convolutional Neural Network
• Multi-View Product Image Search Using Deep ConvNets Representations
• Holistic generational offsets: Fostering a primitive online abstraction for human vs. machine cognition
• Seeing Eye to AI? Applying Deep-Feature-Based Similarity Metrics to Information Visualization
• clcNet: Improving the Efficiency of Convolutional Neural Network using Channel Local Convolutions
• A Brief Survey and an Application of Semantic Image Segmentation for Autonomous Driving
• Efficient Fusion of Sparse and Complementary Convolutions
• Deep Photovoltaic Nowcasting
• Auto Deep Compression by Reinforcement Learning Based Actor-Critic Structure