Search

There is a growing interest in enhancing compiler optimizations with ML models, yet interactions between compilers and ML frameworks remain challenging. Some optimizations require tightly coupled models and compiler internals,raising issues with modularity, performance and framework independence. Practical deployment and transparency for the end-user are also important concerns. We propose ML-Compiler-Bridge to enable ML model development within a traditional Python framework while making end-to-end integration with an optimizing compiler possible and efficient. We evaluate it on both research and production use cases, for training and inference, over several optimization problems, multiple compilers and its versions, and gym infrastructures.

We explore what we call ``omitted label contexts,'' in which training data is limited to a subset of the possible labels. This setting is common among specialized human experts or specific focused studies. We lean on well-studied paradoxes (Simpson's and Condorcet) to illustrate the more general difficulties of causal inference in omitted label contexts. Contrary to the fundamental principles on which much of causal inference is built, we show that ``correct'' adjustments sometimes require non-exchangeable treatment and control groups. These pitfalls lead us to the study networks of conclusions drawn from different contexts and the structures the form, proving an interesting connection between these networks and social choice theory.

Predictions are made for the jet substructure of one-jet events produced in electron-proton neutral current deep inelastic scattering at the future Electron-Ion Collider for exchanged four-momentum squared, $Q^2 > 125$ GeV$^2$. Data are simulated using Monte Carlo event generators PYTHIA 8.304 and RAPGAP 3.308 at the center of mass energies $\sqrt{s}$ = 63.2, 104.9 and 141 GeV. Jets and subjets are produced by longitudinally invariant $k_T$ cluster algorithm. The subjet multiplicity distributions and differential jet shapes are measured for different jet sizes and varying jet-resolution parameter. A comparison is presented between the $k_T$ and anti-$k_T$ cluster algorithms for the study of jets and subjets using the simulated data and the HERA data., Comment: 11 pages, 15 figures

Transport of particles through channels is of paramount importance in physics, chemistry and surface science due to its broad real world applications. Much insights can be gained by observing the transition paths of a particle through a channel and collecting statistics on the lifetimes in the channel or the escape probabilities from the channel. In this paper, we consider the diffusive transport through a narrow conical channel of a Brownian particle subject to intermittent dynamics, namely, stochastic resetting. As such, resetting brings the particle back to a desired location from where it resumes its diffusive phase. To this end, we extend the Fick-Jacobs theory of channel-facilitated diffusive transport to resetting-induced transport. Exact expressions for the conditional mean first passage times, escape probabilities and the total average lifetime in the channel are obtained, and their behaviour as a function of the resetting rate are highlighted. It is shown that resetting can expedite the transport through the channel -- rigorous constraints for such conditions are then illustrated. Furthermore, we observe that a carefully chosen resetting rate can render the average lifetime of the particle inside the channel minimal. Interestingly, the optimal rate undergoes continuous and discontinuous transitions as some relevant system parameters are varied. The validity of our one-dimensional analysis and the corresponding theoretical predictions are supported by three-dimensional Brownian dynamics simulations. We thus believe that resetting can be useful to facilitate particle transport across biological membranes -- a phenomena that can spearhead further theoretical and experimental studies.

We aim to automate decades of research and experience in register allocation, leveraging machine learning. We tackle this problem by embedding a multi-agent reinforcement learning algorithm within LLVM, training it with the state of the art techniques. We formalize the constraints that precisely define the problem for a given instruction-set architecture, while ensuring that the generated code preserves semantic correctness. We also develop a gRPC based framework providing a modular and efficient compiler interface for training and inference. Our approach is architecture independent: we show experimental results targeting Intel x86 and ARM AArch64. Our results match or out-perform the heavily tuned, production-grade register allocators of LLVM., Comment: Published in CC'23

Consider multiple experts with overlapping expertise working on a classification problem under uncertain input. What constitutes a consistent set of opinions? How can we predict the opinions of experts on missing sub-domains? In this paper, we define a framework of to analyze this problem, termed "expert graphs." In an expert graph, vertices represent classes and edges represent binary opinions on the topics of their vertices. We derive necessary conditions for expert graph validity and use them to create "synthetic experts" which describe opinions consistent with the observed opinions of other experts. We show this framework to be equivalent to the well-studied linear ordering polytope. We show our conditions are not sufficient for describing all expert graphs on cliques, but are sufficient for cycles., Comment: 13 pages, 11 figures

Varying domains and biased datasets can lead to differences between the training and the target distributions, known as covariate shift. Current approaches for alleviating this often rely on estimating the ratio of training and target probability density functions. These techniques require parameter tuning and can be unstable across different datasets. We present a new method for handling covariate shift using the empirical cumulative distribution function estimates of the target distribution by a rigorous generalization of a recent idea proposed by Vapnik and Izmailov. Further, we show experimentally that our method is more robust in its predictions, is not reliant on parameter tuning and shows similar classification performance compared to the current state-of-the-art techniques on synthetic and real datasets., Comment: Accepted in Neurips 2020

A method for encoding information in DNA sequences is described. The method is based on the precision-resolution framework, and is aimed to work in conjunction with a recently suggested terminator-free template independent DNA synthesis method. The suggested method optimizes the amount of information bits per synthesis time unit, namely, the writing rate. Additionally, the encoding scheme studied here takes into account the existence of multiple copies of the DNA sequence, which are independently distorted. Finally, quantizers for various run-length distributions are designed., Comment: To appear in ISIT 2020

Deep Neural Networks (DNNs) are a revolutionary force in the ongoing information revolution, and yet their intrinsic properties remain a mystery. In particular, it is widely known that DNNs are highly sensitive to noise, whether adversarial or random. This poses a fundamental challenge for hardware implementations of DNNs, and for their deployment in critical applications such as autonomous driving. In this paper we construct robust DNNs via error correcting codes. By our approach, either the data or internal layers of the DNN are coded with error correcting codes, and successful computation under noise is guaranteed. Since DNNs can be seen as a layered concatenation of classification tasks, our research begins with the core task of classifying noisy coded inputs, and progresses towards robust DNNs. We focus on binary data and linear codes. Our main result is that the prevalent parity code can guarantee robustness for a large family of DNNs, which includes the recently popularized binarized neural networks. Further, we show that the coded classification problem has a deep connection to Fourier analysis of Boolean functions. In contrast to existing solutions in the literature, our results do not rely on altering the training process of the DNN, and provide mathematically rigorous guarantees rather than experimental evidence., Comment: To appear in ISIT '20

Data is one of the most important assets of the information age, and its societal impact is undisputed. Yet, rigorous methods of assessing the quality of data are lacking. In this paper, we propose a formal definition for the quality of a given dataset. We assess a dataset's quality by a quantity we call the expected diameter, which measures the expected disagreement between two randomly chosen hypotheses that explain it, and has recently found applications in active learning. We focus on Boolean hyperplanes, and utilize a collection of Fourier analytic, algebraic, and probabilistic methods to come up with theoretical guarantees and practical solutions for the computation of the expected diameter. We also study the behaviour of the expected diameter on algebraically structured datasets, conduct experiments that validate this notion of quality, and demonstrate the feasibility of our techniques.