arXiv subjects

Michael Carbin

Publications and source records attributed to Michael Carbin.

47 records · Page 3Linked to original sources

arXiv2020

Linear Mode Connectivity and the Lottery Ticket Hypothesis

We study whether a neural network optimizes to the same, linearly connected minimum under different samples of SGD noise (e.g., random data order and augmentation). We find that standard vision models become stable to SGD noise in this way early in training. From then on, the outcome of optimization is determined to a linearly connected region. We use this technique to study iterative magnitude pruning (IMP), the procedure used by work on the lottery ticket hypothesis to identify subnetworks that could have trained in isolation to full accuracy. We find that these subnetworks only reach full accuracy when they are stable to SGD noise, which either occurs at initialization for small-scale settings (MNIST) or early in training for large-scale settings (ResNet-50 and Inception-v3 on ImageNet).

cs.LG↗

arXiv2020

TIRAMISU: A Polyhedral Compiler for Dense and Sparse Deep Learning

In this paper, we demonstrate a compiler that can optimize sparse and recurrent neural networks, both of which are currently outside of the scope of existing neural network compilers (sparse neural networks here stand for networks that can be accelerated with sparse tensor algebra techniques). Our demonstration includes a mapping of sparse and recurrent neural networks to the polyhedral model along with an implementation of our approach in TIRAMISU, our state-of-the-art polyhedral compiler. We evaluate our approach on a set of deep learning benchmarks and compare our results with hand-optimized industrial libraries. Our results show that our approach at least matches Intel MKL-DNN and in some cases outperforms it by 5x (on multicore-CPUs).

cs.DC↗

arXiv2020

Reactive Probabilistic Programming

Synchronous modeling is at the heart of programming languages like Lustre, Esterel, or Scade used routinely for implementing safety critical control software, e.g., fly-by-wire and engine control in planes. However, to date these languages have had limited modern support for modeling uncertainty -- probabilistic aspects of software's environment or behavior -- even though modeling uncertainty is a primary activity when designing a control system. In this paper we present ProbZelus the first synchronous probabilistic programming language. ProbZelus conservatively provides the facilities of a synchronous language to write control software, with probabilistic constructs to model uncertainties and perform inference-in-the-loop. We present the design and implementation of the language. We propose a measure-theoretic semantics of probabilistic stream functions and a simple type discipline to separate deterministic and probabilistic expressions. We demonstrate a semantics-preserving compilation into a first-order functional language that lends itself to a simple presentation of inference algorithms for streaming models. We also redesign the delayed sampling inference algorithm to provide efficient streaming inference. Together with an evaluation on several reactive applications, our results demonstrate that ProbZelus enables the design of reactive probabilistic applications and efficient, bounded memory inference.

cs.PL↗

arXiv2020

Comparing Rewinding and Fine-tuning in Neural Network Pruning

Many neural network pruning algorithms proceed in three steps: train the network to completion, remove unwanted structure to compress the network, and retrain the remaining structure to recover lost accuracy. The standard retraining technique, fine-tuning, trains the unpruned weights from their final trained values using a small fixed learning rate. In this paper, we compare fine-tuning to alternative retraining techniques. Weight rewinding (as proposed by Frankle et al., (2019)), rewinds unpruned weights to their values from earlier in training and retrains them from there using the original training schedule. Learning rate rewinding (which we propose) trains the unpruned weights from their final values using the same learning rate schedule as weight rewinding. Both rewinding techniques outperform fine-tuning, forming the basis of a network-agnostic pruning algorithm that matches the accuracy and compression ratios of several more network-specific state-of-the-art techniques.

cs.LG↗

arXiv2019

Ithemal: Accurate, Portable and Fast Basic Block Throughput Estimation using Deep Neural Networks

Predicting the number of clock cycles a processor takes to execute a block of assembly instructions in steady state (the throughput) is important for both compiler designers and performance engineers. Building an analytical model to do so is especially complicated in modern x86-64 Complex Instruction Set Computer (CISC) machines with sophisticated processor microarchitectures in that it is tedious, error prone, and must be performed from scratch for each processor generation. In this paper we present Ithemal, the first tool which learns to predict the throughput of a set of instructions. Ithemal uses a hierarchical LSTM--based approach to predict throughput based on the opcodes and operands of instructions in a basic block. We show that Ithemal is more accurate than state-of-the-art hand-written tools currently used in compiler backends and static machine code analyzers. In particular, our model has less than half the error of state-of-the-art analytical models (LLVM's llvm-mca and Intel's IACA). Ithemal is also able to predict these throughput values just as fast as the aforementioned tools, and is easily ported across a variety of processor microarchitectures with minimal developer effort.

cs.DC↗

arXiv2019

The Lottery Ticket Hypothesis: Finding Sparse, Trainable Neural Networks

Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of inference without compromising accuracy. However, contemporary experience is that the sparse architectures produced by pruning are difficult to train from the start, which would similarly improve training performance. We find that a standard pruning technique naturally uncovers subnetworks whose initializations made them capable of training effectively. Based on these results, we articulate the "lottery ticket hypothesis:" dense, randomly-initialized, feed-forward networks contain subnetworks ("winning tickets") that - when trained in isolation - reach test accuracy comparable to the original network in a similar number of iterations. The winning tickets we find have won the initialization lottery: their connections have initial weights that make training particularly effective. We present an algorithm to identify winning tickets and a series of experiments that support the lottery ticket hypothesis and the importance of these fortuitous initializations. We consistently find winning tickets that are less than 10-20% of the size of several fully-connected and convolutional feed-forward architectures for MNIST and CIFAR10. Above this size, the winning tickets that we find learn faster than the original network and reach higher test accuracy.

cs.LG↗

arXiv2018

Exploiting Errors for Efficiency: A Survey from Circuits to Algorithms

When a computational task tolerates a relaxation of its specification or when an algorithm tolerates the effects of noise in its execution, hardware, programming languages, and system software can trade deviations from correct behavior for lower resource usage. We present, for the first time, a synthesis of research results on computing systems that only make as many errors as their users can tolerate, from across the disciplines of computer aided design of circuits, digital system design, computer architecture, programming languages, operating systems, and information theory. Rather than over-provisioning resources at each layer to avoid errors, it can be more efficient to exploit the masking of errors occurring at one layer which can prevent them from propagating to a higher layer. We survey tradeoffs for individual layers of computing systems from the circuit level to the operating system level and illustrate the potential benefits of end-to-end approaches using two illustrative examples. To tie together the survey, we present a consistent formalization of terminology, across the layers, which does not significantly deviate from the terminology traditionally used by research communities in their layer of focus.

cs.AR↗

arXiv2018

Verifying Programs Under Custom Application-Specific Execution Models

Researchers have recently designed a number of application-specific fault tolerance mechanisms that enable applications to either be naturally resilient to errors or include additional detection and correction steps that can bring the overall execution of an application back into an envelope for which an acceptable execution is eventually guaranteed. A major challenge to building an application that leverages these mechanisms, however, is to verify that the implementation satisfies the basic invariants that these mechanisms require--given a model of how faults may manifest during the application's execution. To this end we present Leto, an SMT based automatic verification system that enables developers to verify their applications with respect to a first-class execution model specification. Namely, Leto enables software and platform developers to programmatically specify the execution semantics of the underlying hardware system as well as verify assertions about the behavior of the application's resulting execution. In this paper, we present the Leto programming language and its corresponding verification system. We also demonstrate Leto on several applications that leverage application-specific fault tolerance mechanisms.

cs.PL↗

arXiv2018

Verifying Handcoded Probabilistic Inference Procedures

Researchers have recently proposed several systems that ease the process of performing Bayesian probabilistic inference. These include systems for automatic inference algorithm synthesis as well as stronger abstractions for manual algorithm development. However, existing systems whose performance relies on the developer manually constructing a part of the inference algorithm have limited support for reasoning about the correctness of the resulting algorithm. In this paper, we present Shuffle, a programming language for manually developing inference procedures that 1) enforces the basic rules of probability theory, 2) enforces the statistical dependencies of the algorithm's corresponding probabilistic model, and 3) generates an optimized implementation. We have used Shuffle to develop inference algorithms for several standard probabilistic models. Our results demonstrate that Shuffle enables a developer to deliver correct and performant implementations of these algorithms.

cs.PL↗

arXiv2018

Computable decision making on the reals and other spaces via partiality and nondeterminism

Though many safety-critical software systems use floating point to represent real-world input and output, programmers usually have idealized versions in mind that compute with real numbers. Significant deviations from the ideal can cause errors and jeopardize safety. Some programming systems implement exact real arithmetic, which resolves this matter but complicates others, such as decision making. In these systems, it is impossible to compute (total and deterministic) discrete decisions based on connected spaces such as $\mathbb{R}$. We present programming-language semantics based on constructive topology with variants allowing nondeterminism and/or partiality. Either nondeterminism or partiality suffices to allow computable decision making on connected spaces such as $\mathbb{R}$. We then introduce pattern matching on spaces, a language construct for creating programs on spaces, generalizing pattern matching in functional programming, where patterns need not represent decidable predicates and also may overlap or be inexhaustive, giving rise to nondeterminism or partiality, respectively. Nondeterminism and/or partiality also yield formal logics for constructing approximate decision procedures. We implemented these constructs in the Marshall language for exact real arithmetic.

cs.LO↗

arXiv2012

Cryptographic Path Hardening: Hiding Vulnerabilities in Software through Cryptography

We propose a novel approach to improving software security called Cryptographic Path Hardening, which is aimed at hiding security vulnerabilities in software from attackers through the use of provably secure and obfuscated cryptographic devices to harden paths in programs. By "harden" we mean that certain error-checking if-conditionals in a given program P are replaced by equivalent" we mean that adversaries cannot use semi-automatic program analysis techniques to reason about the hardened program paths and thus cannot discover as-yet-unknown errors along those paths, except perhaps through black-box dictionary attacks or random testing (which we can never prevent). Other than these unpreventable attack methods, we can make program analysis aimed at error-finding "provably hard" for a resource-bounded attacker, in the same sense that cryptographic schemes are hard to break. Unlike security-through-obscurity, in Cryptographic Path Hardening we use provably-secure crypto devices to hide errors and our mathematical arguments of security are the same as the standard ones used in cryptography. One application of Cryptographic Path Hardening is that software patches or filters often reveal enough information to an attacker that they can be used to construct error-revealing inputs to exploit an unpatched version of the program. By "hardening" the patch we make it difficult for the attacker to analyze the patched program to construct error-revealing inputs, and thus prevent him from potentially constructing exploits.

cs.SE↗