Publications

This paper studies the problem of globally optimizing a variable of interest that is part of a causal model in which a sequence of interventions can be performed. This problem arises in biology, operational research, communications and, more generally, in all fields where the goal is to optimize an output metric of a system of interconnected nodes. Our approach combines ideas from causal inference, uncertainty

Transformers (Vaswani et al., 2017) have gradually become a key component for many state-of-the-art natural-language-representation models. A recent Transformer-based model — BERT (Devlin et al., 2018) — achieved state-of-the-art results on various natural-language-processing tasks, including GLUE, SQuAD v1.1, and SQuAD v2.0. This model however is computationally prohibitive and has a huge number of parameters

Large transformer-based language models have been shown to be very effective in many classification tasks. However, their computational complexity prevents their use in applications requiring the classification of a large set of candidates. While previous works have investigated approaches to reduce model size, relatively little attention has been paid to techniques to improve batch throughput during inference

In many languages like Arabic, diacritics are used to specify pronunciations as well as meanings. Such diacritics are often omitted in written text, increasing the number of possible pronunciations and meanings for a word. This results in a more ambiguous text making computational processing on such text more difficult. Diacritic restoration is the task of restoring missing diacritics in the written text

We propose a reward function estimation framework for inverse reinforcement learning with deep energy-based policies. We name our method PQR, as it sequentially estimates the Policy, the Qfunction, and the Reward function. PQR does not assume that the reward solely depends on the state, instead it allows for a dependency on the choice of action. Moreover, PQR allows for stochastic state transitions. To

We consider simulating quantum systems on digital quantum computers. We show that the performance of quantum simulation can be improved by simultaneously exploiting the commutativity of Hamiltonian, the sparsity of interactions, and the prior knowledge of initial state. We achieve this using Trotterization for a class of interacting electrons that encompasses various physical systems, including the plane-wave-basis

We study whether one can write a Matrix Product Density Operator (MPDO) as the Gibbs state of a quasi-local parent Hamiltonian. We conjecture this is the case for generic MPDO and give supporting evidences. To investigate the locality of the parent Hamiltonian, we take the approach of checking whether the quantum conditional mutual information decays exponentially. The MPDO we consider are constructed from

Quantum state tomography is a powerful, but resource-intensive, general solution for numerous quantum information processing tasks. This motivates the design of robust tomography procedures that use relevant resources as sparingly as possible. Important cost factors include the number of state copies and measurement settings, as well as classical postprocessing time and memory. In this work, we present

We show that a variant of the surface code — the XZZX code — offers remarkable performance for fault-tolerant quantum computation. The error threshold of this code achieves the zero-rate hashing bound for every single-qubit Pauli noise channel; it is the first explicit code shown to have this universal property. We present numerical evidence that this threshold even exceeds the hashing bound for an experimentally

An invaluable method for probing the physics of a quantum many-body spin system is a mapping to non-interacting effective fermions. We find such mappings using only the frustration graph G of a Hamiltonian H, i.e., the network of anti-commutation relations between the Pauli terms in H in a given basis. We prove theorems based solely on the graph G to identify when H is a free-fermion-solvable model, even

Precipitation nowcasting refers to the forecasting of rainfall and other types of precipitation up to 6 hours ahead (as defined by the World Meteorological Organization)1. Since rainfall can be localized and highly changeable, users of precipitation nowcast typically demand to know the exact time, location and intensity of rainfall. It is therefore necessary to make very high resolution, both spatially

The recent emergence of multilingual pretraining language model (mPLM) has enabled breakthroughs on various downstream crosslingual transfer (CLT) tasks. However, mPLMbased methods usually involve two problems: (1) simply fine-tuning may not adapt generalpurpose multilingual representations to be task-aware on low-resource languages; (2) ignore how cross-lingual adaptation happens for downstream tasks.

Recent work has shown that pre-trained language models such as BERT improve robustness to spurious correlations in the dataset. Intrigued by these results, we find that the key to their success is generalization from a small amount of counter examples where the spurious correlations do not hold. When such minority examples are scarce, pre-trained models perform as poorly as models trained from scratch.

Dialogue response generation models that use template ranking rather than direct sequence generation allow model developers to limit generated responses to pre-approved messages. However, manually creating templates is time consuming and requires domain expertise. To alleviate this problem, we explore automating the process of creating dialogue templates by using unsupervised methods to cluster historical

Electronic records contain sequences of events, some of which take place all at once in a single visit, and others that are dispersed over multiple visits, each with a different timestamp. We postulate that fine temporal detail, e.g., whether a series of blood tests are completed at once or in rapid succession should not alter predictions based on this data. Motivated by this intuition, we propose models