Motivated by problems arising in digital advertising, we introduce the task of training differentially private (DP) machine learning models with semi-sensitive features. In this setting, a subset of the features is known to the attacker (and thus need not be protected) while the remaining features as well as the label are unknown to the attacker and should be protected by the DP guarantee. This task interpolates between training the model with full DP (where the label and all features should be protected) or with label DP (where all the features are considered known, and only the label should be protected). We present a new algorithm for training DP models with semi-sensitive features. Through an empirical evaluation on real ads datasets, we demonstrate that our algorithm surpasses in utility the baselines of (i) DP stochastic gradient descent (DP-SGD) run on all features (known and unknown), and (ii) a label DP algorithm run only on the known features (while discarding the unknown ones).

A well-known algorithm in privacy-preserving ML is differentially private stochastic gradient descent (DP-SGD). While this algorithm has been evaluated on text and image data, it has not been previously applied to ads data, which are notorious for their high class imbalance and sparse gradient updates. In this work we apply DP-SGD to several ad modeling tasks including predicting click-through rates, conversion rates, and number of conversion events, and evaluate their privacy-utility trade-off on real-world datasets. Our work is the first to empirically demonstrate that DP-SGD can provide both privacy and utility for ad modeling tasks.

This work presents Origami, a framework which provides privacy-preserving inference for large deep neural network (DNN) models through a combination of enclave execution, cryptographic blinding, interspersed with accelerator-based computation. Origami partitions the ML model into multiple partitions. The first partition receives the encrypted user input within an SGX enclave. The enclave decrypts the input and then applies cryptographic blinding to the input data and the model parameters. The layer computation is offloaded to a GPU/CPU and the computed output is returned to the enclave, which decodes the computation on noisy data using the unblinding factors privately stored within SGX. This process may be repeated for each DNN layer, as has been done in prior work Slalom. However, the overhead of blinding and unblinding the data is a limiting factor to scalability. Origami relies on the empirical observation that the feature maps after the first several layers can not be used, even by a powerful conditional GAN adversary to reconstruct input. Hence, Origami dynamically switches to executing the rest of the DNN layers directly on an accelerator. We empirically demonstrate that using Origami, a conditional GAN adversary, even with an unlimited inference budget, cannot reconstruct the input. Compared to running the entire VGG-19 model within SGX, Origami inference improves the performance of private inference from 11x while using Slalom to 15. 1x.

MLaaS (ML-as-a-Service) offerings by cloud computing platforms are becoming increasingly popular. Hosting pre-trained machine learning models in the cloud enables elastic scalability as the demand grows. But providing low latency and reducing the latency variance is a key requirement. Variance is harder to control in a cloud deployment due to uncertainties in resource allocations across many virtual instances. We propose the collage inference technique, which uses a novel convolutional neural network model, collage-cnn, to provide low-cost redundancy. A collage-cnn model takes a collage image formed by combining multiple images and performs multi-image classification in one shot, albeit at slightly lower accuracy. We augment a collection of traditional single image classifier models with a single collage-cnn classifier, which acts as their low-cost redundant backup. Collage-cnn provides backup classification results if any single image classification requests experience a slowdown. Deploying the collage-cnn models in the cloud, we demonstrate that the 99th percentile tail latency of inference can be reduced by 1.2x to 2x compared to replication-based approaches while providing high accuracy. Variation in inference latency can be reduced by 1.8x to 15x.

MLaaS (ML-as-a-Service) offerings by cloud computing platforms are becoming increasingly popular these days. Pre-trained machine learning models are deployed on the cloud to support prediction based applications and services. For achieving higher throughput, incoming requests are served by running multiple replicas of the model on different machines concurrently. Incidence of straggler nodes in distributed inference is a significant concern since it can increase inference latency, violate SLOs of the service. In this paper, we propose a novel coded inference model to deal with stragglers in distributed image classification. We propose modified single shot object detection models, Collage-CNN models, to provide necessary resilience efficiently. A Collage-CNN model takes collage images formed by combining multiple images as its input and performs multi-image classification in one shot. We generate custom training collages using images from standard image classification datasets and train the model to achieve high classification accuracy. Deploying the Collage-CNN models in the cloud, we demonstrate that the 99th percentile latency can be reduced by 1.45X to 2.46X compared to replication based approaches and without compromising prediction accuracy.

MLaaS (ML-as-a-Service) offerings by cloud computing platforms are becoming increasingly popular. Hosting pre-trained machine learning models in the cloud enables elastic scalability as the demand grows. But providing low latency and reducing the latency variance is a key requirement. Variance is harder to control in a cloud deployment due to uncertainties in resource allocations across many virtual instances. We propose the collage inference technique which uses a novel convolutional neural network model, collage-cnn, to provide low-cost redundancy. A collage-cnn model takes a collage image formed by combining multiple images and performs multi-image classification in one shot, albeit at slightly lower accuracy. We augment a collection of traditional single image classifier models with a single collage-cnn classifier which acts as their low-cost redundant backup. Collage-cnn provides backup classification results if any single image classification requests experience slowdown. Deploying the collage-cnn models in the cloud, we demonstrate that the 99th percentile tail latency of inference can be reduced by 1.2x to 2x compared to replication based approaches while providing high accuracy. Variation in inference latency can be reduced by 1.8x to 15x.

Reducing the latency variance in machine learning inference is a key requirement in many applications. Variance is harder to control in a cloud deployment in the presence of stragglers. In spite of this challenge, inference is increasingly being done in the cloud, due to the advent of affordable machine learning as a service (MLaaS) platforms. Existing approaches to reduce variance rely on replication which is expensive and partially negates the affordability of MLaaS. In this work, we argue that MLaaS platforms also provide unique opportunities to cut the cost of redundancy. In MLaaS platforms, multiple inference requests are concurrently received by a load balancer which can then create a more cost-efficient redundancy coding across a larger collection of images. We propose a novel convolutional neural network model, Collage-CNN, to provide a low-cost redundancy framework. A Collage-CNN model takes a collage formed by combining multiple images and performs multi-image classification in one shot, albeit at slightly lower accuracy. We then augment a collection of traditional single image classifiers with a single Collage-CNN classifier which acts as a low-cost redundant backup. Collage-CNN then provides backup classification results if a single image classification straggles. Deploying the Collage-CNN models in the cloud, we demonstrate that the 99th percentile tail latency of inference can be reduced by 1.47X compared to replication based approaches while providing high accuracy. Also, variation in inference latency can be reduced by 9X with a slight increase in average inference latency.

While performing distributed computations in today's cloud-based platforms, execution speed variations among compute nodes can significantly reduce the performance and create bottlenecks like stragglers. Coded computation techniques leverage coding theory to inject computational redundancy and mitigate stragglers in distributed computations. In this paper, we propose a dynamic workload distribution strategy for coded computation called Slack Squeeze Coded Computation ( S 2 C 2 ). S 2 C 2 squeezes the compute slack (i.e., overhead) that is built into the coded computing frameworks by efficiently assigning work for all fast and slow nodes according to their speeds and without needing to re-distribute data. We implement an LSTM-based speed prediction algorithm to predict speeds of compute nodes. We evaluate S 2 C 2 on linear algebraic algorithms, gradient descent, graph ranking, and graph filtering algorithms. We demonstrate 19% to 39% reduction in total computation latency using S 2 C 2 compared to job replication and coded computation. We further show how S 2 C 2 can be applied beyond matrix-vector multiplication.

This work presents Origami, which provides privacy-preserving inference for large deep neural network (DNN) models through a combination of enclave execution, cryptographic blinding, interspersed with accelerator-based computation. Origami partitions the ML model into multiple partitions. The first partition receives the encrypted user input within an SGX enclave. The enclave decrypts the input and then applies cryptographic blinding to the input data and the model parameters. Cryptographic blinding is a technique that adds noise to obfuscate data. Origami sends the obfuscated data for computation to an untrusted GPU/CPU. The blinding and de-blinding factors are kept private by the SGX enclave, thereby preventing any adversary from denoising the data, when the computation is offloaded to a GPU/CPU. The computed output is returned to the enclave, which decodes the computation on noisy data using the unblinding factors privately stored within SGX. This process may be repeated for each DNN layer, as has been done in prior work Slalom. However, the overhead of blinding and unblinding the data is a limiting factor to scalability. Origami relies on the empirical observation that the feature maps after the first several layers can not be used, even by a powerful conditional GAN adversary to reconstruct input. Hence, Origami dynamically switches to executing the rest of the DNN layers directly on an accelerator without needing any further cryptographic blinding intervention to preserve privacy. We empirically demonstrate that using Origami, a conditional GAN adversary, even with an unlimited inference budget, cannot reconstruct the input. We implement and demonstrate the performance gains of Origami using the VGG-16 and VGG-19 models. Compared to running the entire VGG-19 model within SGX, Origami inference improves the performance of private inference from 11x while using Slalom to 15.1x.

Training a machine learning model is both compute and data-intensive. Most of the model training is performed on high performance compute nodes and the training data is stored near these nodes for faster training. But there is a growing interest in enabling training near the data. For instance, mobile devices are rich sources of training data. It may not be feasible to consolidate the data from mobile devices into a cloud service, due to bandwidth and data privacy reasons. Training at mobile devices is however fraught with challenges. First mobile devices may join or leave the distributed setting, either voluntarily or due to environmental uncertainties, such as lack of power. Tolerating uncertainties is critical to the success of distributed mobile training. One proactive approach to tolerate computational uncertainty is to store data in a coded format and perform training on coded data. Encoding data is a challenging task since erasure codes require multiple devices to exchange their data to create a coded data partition, which places a significant bandwidth constraint. Furthermore, coded computing traditionally relied on a central node to encode and distribute data to all the worker nodes, which is not practical in a distributed mobile setting. In this paper, we tackle the uncertainty in distributed mobile training using a bandwidth-efficient encoding strategy. We use a Random Linear Network coding (RLNC) which reduces the need to exchange data partitions across all participating mobile devices, while at the same time preserving the property of coded computing to tolerate uncertainties. We implement gradient descent for logistic regression and SVM to evaluate the effectiveness of our mobile training framework. We demonstrate a 50% reduction in total required communication bandwidth compared to MDS coded computation, one of the popular erasure codes.

Data parallelism can boost the training speed of convolutional neural networks (CNN), but could suffer from significant communication costs caused by gradient aggregation. To alleviate this problem, several scalar quantization techniques have been developed to compress the gradients. But these techniques could perform poorly when used together with decentralized aggregation protocols like ring all-reduce (RAR), mainly due to their inability to directly aggregate compressed gradients. In this paper, we empirically demonstrate the strong linear correlations between CNN gradients, and propose a gradient vector quantization technique, named GradiVeQ, to exploit these correlations through principal component analysis (PCA) for substantial gradient dimension reduction. GradiVeQ enables direct aggregation of compressed gradients, hence allows us to build a distributed learning system that parallelizes GradiVeQ gradient compression and RAR communications. Extensive experiments on popular CNNs demonstrate that applying GradiVeQ slashes the wall-clock gradient aggregation time of the original RAR by more than 5X without noticeable accuracy loss, and reduces the end-to-end training time by almost 50%. The results also show that GradiVeQ is compatible with scalar quantization techniques such as QSGD (Quantized SGD), and achieves a much higher speed-up gain under the same compression ratio.

The adoption of SSDs has become very prominent, but they still suffer from challenges to control write amplification. Traditional SSDs have single active append point where new data writes can be stored. Data of different lifetime stored together causes high write amplification. Recently, multi-stream SSDs are developed that allows multiple active append points. These multiple active append points can be used to store data of different lifetime in different locations within SSD. Such a data placement according to the lifetime of data would considerably reduce internal write amplification of SSD. For using multi-stream SSDs it is required to attach stream-id to each new incoming data writes. According to these stream-ids, the flash transition layer (FTL) of a multi-stream SSD then appends data to different erase blocks. Thus, multi-stream SSDs will help to reduce write amplification. But, to efficiently use this new multi-stream SSDs, it is important to properly identify streamids of data with respect to its lifetime. The lifetime of data is expected using different features that data exhibits like frequency, sequentiality etc. Stream-id identification using different features may have different impact on the final write amplification of multi-stream SSDs, depending on workload. Thus, it is required to quantify the impact of different data features that are used for stream-id identification on the resultant write amplification. Additionally, the combination of these data features may be used for stream-id identification, so it is also important to be able to study the impact of such different combinations. In order to address above challenges towards efficiently using multi-stream SSDs, here we propose a portable and adoptable framework to study the impact of stream-id identification using different workload data features and their combinations on write amplification of multi-stream SSDs. Our evaluation results show that use of appropriate features according to workload can considerably reduce the Write Amplification Factor (WAF) when compared to the legacy SSDs.

Modern data center solid state drives (SSDs) integrate multiple general-purpose embedded cores to manage ash translation layer, garbage collection, wear-leveling, and etc., to improve the performance and the reliability of SSDs. As the performance of these cores steadily improves there are opportunities to repurpose these cores to perform application driven computations on stored data, with the aim of reducing the communication between the host processor and the SSD. Reducing host-SSD bandwidth demand cuts down the I/O time which is a bottleneck for many applications operating on large data sets. However, the embedded core performance is still significantly lower than the host processor, as generally wimpy embedded cores are used within SSD for cost effective reasons. So there is a trade-o between the computation overhead associated with near SSD processing and the reduction in communication overhead to the host system. In this work, we design a set of application programming interfaces (APIs) that can be used by the host application to offload a data intensive task to the SSD processor. We describe how these APIs can be implemented by simple modifications to the existing Non-Volatile Memory Express (NVMe) command interface between the host and the SSD processor. We then quantify the computation versus communication tradeoffs for near storage computing using applications from two important domains, namely data analytics and data integration. Using a fully functional SSD evaluation platform we perform design space exploration of our proposed approach by varying the bandwidth and computation capabilities of the SSD processor. We evaluate static and dynamic approaches for dividing the work between the host and SSD processor, and show that our design may improve the performance by up to 20% when compared to processing at the host processor only, and 6when compared to processing at the SSD processor only. CCS CONCEPTS • Computer systems organization → Secondary storage organization; Distributed architectures; Firmware;

Accelerated life testing (ALT) of a field programmable gate array (FPGA) requires it to be configured with a circuit that satisfies multiple criteria. Hand-crafting such a circuit is a herculean task as many components of the criteria are orthogonal to each other demanding a complex multivariate optimization. This paper presents an evolutionary algorithm aided by particle swarm optimization methodology to generate synthetic benchmark circuits (SBC) that can be used for ALT of FPGAs. The proposed algorithm was used to generate a SBC for ALT of a commercial FPGA. The generated SBC when compared with a hand-crafted one, demonstrated to be more suitable for ALT, measured in terms of meeting the multiple criteria. The SBC generated by the proposed technique utilizes 8.37% more resources; operates at a maximum frequency which is 40% higher; and has 7.75% higher switching activity than the hand-crafted one reported in the literature. The hand-crafted circuit is very specific to the particular device of that family of FPGAs, whereas the proposed algorithm is device-independent. In addition, it took several man months to hand-craft the SBC, whereas the proposed algorithm took less than half-a-day.