Paper proposes a faster method for evaluating DNN hardware and software designs.
problem Reducing time for evaluating different DNN hardware and software designs.
method Using virtual hardware models to estimate DNN performance at the concept phase.
result Up to 92% accuracy in predicting DNN inference processing time.
MLPerf benchmarks ML inference systems across diverse hardware.
problem Challenges in evaluating ML inference systems due to variety and complexity.
method MLPerf prescribes a set of rules and best practices for comparability.
result First call for submissions yielded over 600 reproducible measurements.
Benchmark for DL inference on embedded HWAs, focusing on autonomous driving.
problem Lack of comprehensive benchmarks for DL hardware.
method Developed a benchmark for inference on embedded HWAs, focusing on autonomous driving. Proposed new granularity, benchmark procedures, and performance indicators.
result Identifies mismatches between HWAs and DL models.
This paper tackles co-design of neural hardware and software to improve efficiency.
problem Designing efficient deep learning systems that consider both hardware and software optimizations together.
method Developed a constrained Bayesian optimization framework to automatically identify profitable design points in the joint hardware/software design space.
result Improved energy-delay product by 18% (ResNet) and 40% (DQN) over hand-tuned systems.
Automated design of resilient, efficient DNNs for hardware.
problem Designing reliable and efficient DNNs for hardware.
method Evolutionary optimization technique for DNN architecture design.
result Strong correlation between predicted and actual error resilience.
Co-designing efficient machine learning based systems across the whole hardware/software stack to trade off speed, accuracy, energy and costs is becoming extremely complex and time consuming. Researchers often struggle to evaluate and compare different published works across rapidly evolving software frameworks, hetero…
New KWS neural networks improve accuracy and power efficiency.
problem Maximizing accuracy and power efficiency in KWS systems.
method Hardware Aware Training (HAT) for LMU-based neural networks.
result Achieved state-of-the-art accuracy and low parameter counts.
Recent breakthroughs in Deep Learning (DL) applications have made DL models a key component in almost every modern computing system. The increased popularity of DL applications deployed on a wide-spectrum of platforms have resulted in a plethora of design challenges related to the constraints introduced by the hardware…
Efficient FPGA dropout algorithm reduces memory usage.
problem Overfitting in deep neural networks.
method Hardware-oriented dropout algorithm for FPGA implementation.
result Significant resource reduction in FPGA implementation.
In this work, a new approach for Sun tracking systems is presented. Due to the current system limitations regarding costs and operational problems, a new approach based on low cost, computer vision open hardware and deep learning has been developed. The preliminary tests carried out successfully in Plataforma solar de …
Method predicts hardware resource usage by control software with guaranteed linear convergence.
problem Predicting time-varying hardware resource availability in control software.
method Path structured multimarginal Schrödinger bridge (MSBP) for learning stochastic resource usage.
result Guaranteed linear convergence to accurate prediction of hardware resource utilization.
We present a design and implementation of the Thomas algorithm optimized for hardware acceleration on an FPGA, the Thomas Core. The hardware-based algorithm combined with the custom data flow and low level parallelism available in an FPGA reduces the overall complexity from 8N down to 5N serial arithmetic operations, a…
This paper highlights new opportunities for designing large-scale machine learning systems as a consequence of blurring traditional boundaries that have allowed algorithm designers and application-level practitioners to stay -- for the most part -- oblivious to the details of the underlying hardware-level implementatio…
Optimized neural networks for Edge TPU achieve high accuracy in real-time image classification.
problem Designing neural networks for hardware accelerators to achieve optimal performance.
method Hardware-aware neural architecture search and model customization for Edge TPU.
result Improved accuracy-latency tradeoff on Pixel 4's Edge TPU compared to existing models.
Hardware accelerations of deep learning systems have been extensively investigated in industry and academia. The aim of this paper is to achieve ultra-high energy efficiency and performance for hardware implementations of deep neural networks (DNNs). An algorithm-hardware co-optimization framework is developed, which i…
Predicting MRI coil failures using time series classification.
problem Early detection of MRI hardware failures to prevent downtime.
method Training a statistical model on sequential image data features over time.
result LSTMs achieved an F-score of 86.43% and 98.33% accuracy in predicting coil damage.
We derive scaling laws for optimizing neural networks in hardware.
problem Optimizing the large parameter space of neural networks in hardware.
method Analytical derivation of scaling laws for Coordinate Descent optimization.
result Convergence is exponential and scales linearly with the number of neurons.
MLSys aims to bridge ML and systems research.
problem Designing ML systems for real-world deployment is challenging.
method Foster a new conference and research community.
result MLSys conference focuses on intersection of systems and ML.
There is a recent interest in neural network (NN)-based communication algorithms which have shown to achieve (beyond) state-of-the-art performance for a variety of problems or lead to reduced implementation complexity. However, most work on this topic is simulation based and implementation on specialized hardware for f…
HarDNN detects and protects CNNs from hardware errors.
problem Hardware errors in CNNs can corrupt inference output.
method Statistical error injection and heuristic vulnerability assessment.
result HarDNN improves CNN resilience with minimal additional computation.
XSP profiles ML models across hardware and software stacks.
problem Challenges in profiling ML model performance across different layers of the stack.
method XSP uses distributed tracing to aggregate data from various sources and introduces a leveled, iterative measurement approach.
result XSP provides insights into ML model execution not easily discernible otherwise.
Survey on deep neural networks for vision and speech.
problem Improving intelligent vision and speech systems.
method Review of deep learning models and challenges.
result Emerging technologies show promise for future systems.
TinyCNN accelerates CNN models on embedded FPGA with 15x speedup.
problem Limited memory on embedded FPGAs restricts CNN performance.
method Software and hardware design tool for FPGA resource-aware CNN accelerator.
result 3% accuracy loss with 15.75x speedup on image classification.
Specialized Deep Learning (DL) acceleration stacks, designed for a specific set of frameworks, model architectures, operators, and data types, offer the allure of high performance while sacrificing flexibility. Changes in algorithms, models, operators, or numerical systems threaten the viability of specialized hardware…
Improves VQAs by balancing classical and quantum training resources.
problem Challenges in trainability and resource costs of VQAs on quantum hardware.
method Adopting HELIA Ansatz and combining classical and quantum methods for gradient estimation and training.
result Achieves higher accuracy and success rates in VQE and improved test accuracy in quantum phase classification.
New method uses surrogate gradients to train efficient spiking networks on neuromorphic hardware.
problem Training high-performing spiking networks on analog neuromorphic hardware is challenging due to device mismatch and lack of efficient algorithms.
method Introduces a general in-the-loop learning framework based on surrogate gradients.
result Learning self-corrects for device mismatch, resulting in competitive spiking network performance.
A scalable DL benchmarking platform for evaluating and comparing models, frameworks, and hardware.
problem Lack of a uniform DL benchmarking platform for evaluating and comparing innovations.
method Identified 10 design features for a DL benchmarking platform, proposed MLModelScope, and implemented as an open-source project.
result Demonstrated how model, hardware, and framework selection affect accuracy and performance under different scenarios.
Despite being originally inspired by the central nervous system, artificial neural networks have diverged from their biological archetypes as they have been remodeled to fit particular tasks. In this paper, we review several possibilites to reverse map these architectures to biologically more realistic spiking networks…
The intrinsic error tolerance of neural network (NN) makes approximate computing a promising technique to improve the energy efficiency of NN inference. Conventional approximate computing focuses on balancing the efficiency-accuracy trade-off for existing pre-trained networks, which can lead to suboptimal solutions. In…
SELD-TCN improves sound event localization and detection efficiency.
problem Efficient sound event localization and detection on embedded hardware.
method Developed a novel temporal convolutional network (TCN) architecture.
result SELD-TCN outperforms state-of-the-art SELDnet on four datasets.
The application of deep learning techniques resulted in remarkable improvement of machine learning models. In this paper provides detailed characterizations of deep learning models used in many Facebook social network services. We present computational characteristics of our models, describe high performance optimizati…
Machine learning improves PMD compensation in multiplexed systems.
problem Improving performance in multiplexed systems with PMD.
method Model-based machine learning parameterizing the Manakov-PMD equation.
result Performance close to PMD-free case achieved with hardware-friendly DBP and PMD compensation.
A new hierarchy quantifies agency in systems based on information processing.
problem Lack of a measurable, universal definition for agency in intelligent systems.
method Developed a bottom-up framework based on information processing hierarchy.
result Identified three orders of information processing (I, II, III) as necessary for agency.
Condensa programmatically optimizes neural network compression.
problem Finding optimal compression strategies for neural networks.
method Bayesian optimization-based algorithm for automatic sparsity inference.
result Significant memory and runtime improvements for real-world DNNs.
SAAD enhances anomaly detection in automotive systems with high accuracy.
problem Anomaly detection in automotive systems.
method Integrates statistical techniques with machine learning.
result SAAD achieves 88.3% accuracy and 0.921 F1 score.
New principle reduces load imbalance in LLM serving systems, saving up to 52% energy.
problem Wasted computational power due to load imbalance in LLM serving systems.
method Developed a universal load-balancing principle for barrier-synchronized systems with non-migratable state.
result Proves worst-case theoretical guarantees for imbalance reduction and energy savings.
Predicting MRI coil failures using image features and ensemble learning.
problem Ensuring trouble-free operation of MRI systems by predicting hardware failures.
method Two-level data analysis with neural networks and Random Forest ensemble learning.
result Improved prediction results with an F-score of 94.14% and an accuracy of 99.09%.
Quantum memory limits set by relativity theory.
problem Quantum memory efficiency and relativity constraints.
method Relativistic quantum field theory and Lieb-Robinson bounds.
result Quantum memory capacity is limited by fundamental physics.
Photonic co-processor speeds up training of large neural networks.
problem Training large neural networks with backpropagation is inefficient and communication is a bottleneck.
method Direct Feedback Alignment (DFA) with a photonic accelerator.
result Photonic accelerator can compute random projections with trillions of parameters.
We introduce a learning-based framework to optimize tensor programs for deep learning workloads. Efficient implementations of tensor operators, such as matrix multiplication and high dimensional convolution, are key enablers of effective deep learning systems. However, existing systems rely on manually optimized librar…
Packed-Ensembles improve uncertainty estimation in constrained hardware.
problem Hardware limitations restrict the size of ensembles and network capacity, degrading performance.
method Packed-Ensembles (PE) design and train lightweight structured ensembles by modulating encoding space and parallelizing into a single backbone.
result PE accurately preserves diversity and maintains performance on key metrics like accuracy, calibration, and out-of-distribution detection.
This dissertation advances scalable Gaussian processes using iterative methods and pathwise conditioning.
problem The classical Gaussian process formulation is not scalable for large datasets and modern hardware.
method Combining iterative methods and pathwise conditioning to improve scalability.
result Significantly reduced memory requirements and facilitated application to larger datasets.
Bio-inspired neuromorphic hardware is a research direction to approach brain's computational power and energy efficiency. Spiking neural networks (SNN) encode information as sparsely distributed spike trains and employ spike-timing-dependent plasticity (STDP) mechanism for learning. Existing hardware implementations of…
DNN is presenting human-level performance for many complex intelligent tasks in real-world applications. However, it also introduces ever-increasing security concerns. For example, the emerging adversarial attacks indicate that even very small and often imperceptible adversarial input perturbations can easily mislead t…
Stealthy hardware Trojan exploits DLA architecture vulnerabilities.
problem Security of DLA deployed on hardware accelerators.
method Input Interception Attack (IIA) exploiting statistical properties of DLA outputs.
result Stealthy Trojan can trigger with some definiteness.
Recurrent neural networks (RNNs) are powerful constructs capable of modeling complex systems, up to and including Turing Machines. However, learning such complex models from finite training sets can be difficult. In this paper we empirically show that RNNs can learn models of computer peripheral devices through input a…
MLPerf benchmarks ML training to drive performance improvements.
problem Unique challenges in ML training benchmarks.
method Developed MLPerf to overcome ML training's specific challenges.
result Quantitatively evaluated MLPerf's effectiveness.
Forward inference techniques such as sequential Monte Carlo and particle Markov chain Monte Carlo for probabilistic programming can be implemented in any programming language by creative use of standardized operating system functionality including processes, forking, mutexes, and shared memory. Exploiting this we have …