Binary and ternary weights simplify RNNs for mobile devices.
problem Complexity and memory intensity of RNNs on mobile devices.
method Learn binary and ternary weights during training.
result Significant memory saving and inference speedup on ASIC platform.
Major advancements in building general-purpose and customized hardware have been one of the key enablers of versatility and pervasiveness of machine learning models such as deep neural networks. To sustain this ubiquitous deployment of machine learning models and cope with their computational and storage complexity, se…
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.
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.
CodeX improves DNN acceleration on FPGAs by encoding and customizing bitwidth.
problem Efficiently accelerating deep neural networks on FPGAs with limited memory.
method Nonlinear encoding, automated bitwidth customization, FPGA streaming buffers.
result Average 4.65x throughput improvement on MNIST, SVHN, CIFAR-10.
New method speeds up sparse graph neural networks training on dense hardware.
problem Training sparse graph neural networks is slow on custom hardware.
method Inspired by sparse matrix optimization, developed techniques for dense hardware.
result Sparse graph neural networks trained in 13 minutes on 512-core TPUv2 Pod.
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…
With ever-increasing computational demand for deep learning, it is critical to investigate the implications of the numeric representation and precision of DNN model weights and activations on computational efficiency. In this work, we explore unconventional narrow-precision floating-point representations as it relates …
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…
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…
Efficient non-uniform quantizer improves CNN performance on FPGA.
problem Improving CNN performance on low-power systems like mobile devices.
method Custom hardware-friendly non-uniform quantizer for parameters and activations using a single scale integer representation.
result Little degradation in accuracy on CIFAR-10 and CIFAR-100 datasets.
The thesis explores how to integrate machine learning with hardware constraints.
problem Designing efficient neural networks for real-time processing with hardware limitations.
method Developed a library for training and converting sparse quantized neural networks to hardware.
result Demonstrated how to design and optimize neural networks for FPGA-based hardware.
OverQ increases model accuracy by handling outliers in neural networks with minimal hardware changes.
problem Handling outliers in neural network weights and activations for low-precision quantization.
method Overwrite quantization (OverQ) that opportunistically increases bitwidth for activation outliers.
result OverQ can handle over 90% of outliers and achieve +5% ImageNet Top-1 accuracy on a quantized ResNet-50 at 4 bits.
Topic models such as Latent Dirichlet Allocation (LDA) have been widely used in information retrieval for tasks ranging from smoothing and feedback methods to tools for exploratory search and discovery. However, classical methods for inferring topic models do not scale up to the massive size of today's publicly availab…
Introduces Quantum Data Center for quantum era benefits.
problem No specific problem stated; focuses on future potential.
method Combines QRAM and quantum networks.
result QDC offers efficiency, security, and precision.
Reservoir computing models classify ECG signals for patient-adaptive monitoring.
problem Classifying ECG signals for patients with imbalanced heartbeat classes.
method Reservoir computing paradigm applied to recurrent neural networks (RNNs).
result Accurate patient-adaptive ECG classifier that handles imbalanced classes.
Hearing Aid (HA) algorithms need to be tuned ("fitted") to match the impairment of each specific patient. The lack of a fundamental HA fitting theory is a strong contributing factor to an unsatisfying sound experience for about 20% of hearing aid patients. This paper proposes a probabilistic modeling approach to the de…
Framework for verifying deep learning operators.
problem Complexity and errors in designing and implementing custom operators.
method Symbolic execution, syntax-guided synthesis, SMT-based verification.
result Effective synthesis and verification of deep learning operators.
New neuromorphic hardware learns MNIST digits efficiently.
problem Limited scalability and in-hardware learning in existing neuromorphic hardware.
method Low-cost scalable NoC-based SNN architecture with in-hardware STDP learning.
result Demonstrated learning capability of the hardware architecture.
The purpose of this study was to build a customer selection model based on 20 dimensions, including customer codes, total contribution, assets, deposit, profit, profit rate, trading volume, trading amount, turnover rate, order amount, withdraw amount, withdraw rate, process fee, process fee submitted, process fee retai…
SmallML predicts customer churn for SMEs with small data, improving accuracy by 24.2 points.
problem AI exclusion of SMEs due to data scale mismatch.
method Bayesian transfer learning with hierarchical pooling and conformal prediction.
result 96.7% AUC on 100 obs SMEs, 24.2 point improvement over logistic regression.
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.
DCE learns customer embeddings from digital activity and financial context.
problem Comprehensive customer understanding in financial services.
method Leverages customers' digital activity and financial context to learn dense representations.
result DCE showed performance lift in three prediction problems.
Reduce survey questions to scale market research without annoying customers.
problem Performing market research by surveying customers with many questions is inefficient and annoying.
method Used Bayesian networks to model and reduce the number of questions asked to customers.
result Demonstrated the effectiveness of the approach using an example of broadband customer segmentation.
Financial institutions use LSTM models to predict customer goals.
problem Predicting customer goals and actions in financial services.
method Used LSTM models with state-space graph embeddings on historical customer traces.
result Demonstrated the effectiveness of LSTM models in predicting customer goals and actions.
Study shows BNN inference accelerators are vulnerable to soft errors, causing significant misclassification.
problem Vulnerability of BNN inference accelerators to soft errors affecting image classification accuracy.
method Fault injection experiments on CIFAR-10 and MNIST datasets to assess impact of soft errors.
result Accuracy of image classifier can drop by up to 76.70% in worst-case scenarios.
Customer momentum is a positive relationship between a firm's returns and past returns of its customers.
problem Understanding the relationship between a firm's returns and its customers' past returns.
method Examined customer momentum using a long-short equally-weighted decile portfolio and Fama-French factor models.
result Customer momentum generates significant monthly returns and is statistically significant.
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.
New Ising models improve consensus clustering on specialized hardware.
problem Consensus clustering optimization problems.
method Formulated consensus clustering as Ising models and evaluated on specialized hardware.
result Our Ising models outperform existing techniques on consensus clustering.
DANCE optimizes neural network and accelerator design for faster, more efficient DNN execution.
problem Challenges in optimizing neural network and accelerator design for efficient DNN execution.
method Differentiable approach to co-exploration of accelerator and network architecture design.
result Significantly shorter time to achieve superior accuracy and hardware cost metrics.
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.
Method generates realistic customer transaction sequences for retail analysis.
problem Modeling customer behavior and purchasing patterns in retail databases.
method Customer embedding using RNN and GAN for generating plausible baskets.
result Generated baskets resemble real transactions and replicate sequential patterns.
Paper proposes a new topology for AML analysis using Poincaré embeddings.
problem Complex money laundering schemes and regulatory constraints hinder AML analysis and information sharing.
method Proposes a new topology for AML analysis using Poincaré embeddings.
result Demonstrates improved AML analysis and information sharing through Poincaré embeddings.
Paper proposes a method to aggregate customer engagement data for better ranking of e-commerce results.
problem Cold start problem and under-representation of new or under-impressed products in e-commerce search results.
method Aggregates customer engagements within a day for the same query as input training data for machine learning models.
result Training models on aggregated data leads to better ranking of new and under-impressed products.
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…
Study uses Open Banking data to estimate customer value, showing potential 21% increase.
problem Limited CLV estimation using single-entity data.
method Introduces PCLV framework using Open Banking data for comprehensive customer value estimation.
result Open Banking data can estimate PCLV per competitor, showing a 21.06% increase over Actual CLV.
Predicts customer call intent for auto dealerships using CNN.
problem Understanding customer intent from phone calls for better service.
method Developed a CNN-based supervised learning model for multi-class classification.
result CNN model performs well on customer call intent classification.
Quantum algorithms for CVaR portfolio optimization face trade-offs between hardware coherence and expressibility.
problem Quantum algorithmic resilience for CVaR portfolio optimization
method WS-QAOA vs. HE-VQNN
result WS-QAOA provides exact theoretical mapping but suffers from hardware decoherence, while HE-VQNN preserves hardware coherence but lacks expressibility.
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.
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.
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.
VAE learns latent representations for bank customers' creditworthiness.
problem Improving marketing, CRM, and credit risk assessment in retail banks.
method Adopted Variational Autoencoder (VAE) to learn latent representations.
result VAE latent representations capture customers' creditworthiness and generalize to new data.
The paper uses RFM and clustering to segment bank customers.
problem Challenges in customer retention and profitable segmentation in banking.
method RFM technique and clustering algorithms applied to real data.
result Successful customer segmentation improves conversion rates.
Design automation optimizes deep learning models for various hardware.
problem Designing efficient deep learning models requires balancing algorithm and hardware.
method Proposes design automation techniques for specialized neural networks, including auto pruning and quantization.
result Learning-based automation achieves superior performance and efficiency compared to human design.
The study improves CLV predictions in retail banking with machine learning.
problem Estimating customer lifetime value in retail banking is challenging.
method Developed a novel framework for CLV predictions over arbitrary time horizons.
result 43% improvement in out-of-time CLV prediction error.
We provide a benchmark dataset for hand gesture recognition using force myography.
problem Lack of publicly available benchmark data for force myography hand gesture recognition.
method Collected data from 20 persons covering 18 unique gestures using a commercially available sensor setup.
result Improved gesture recognition accuracy through transfer learning.
Study clusters bank customers using LSTM and DTW.
problem Efficiently segmenting bank customers for targeted offers.
method Encoder-decoder LSTM network and Dynamic Time Warping (DTW).
result Hybrid method yields more accurate clusters.
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.