1,401 Hits in 1.5 sec

Graph Neural Networks and Boolean Satisfiability [article]

Benedikt Bünz, Matthew Lamm
2017 arXiv   pre-print
In this paper we explore whether or not deep neural architectures can learn to classify Boolean satisfiability (SAT). We devote considerable time to discussing the theoretical properties of SAT. Then, we define a graph representation for Boolean formulas in conjunctive normal form, and train neural classifiers over general graph structures called Graph Neural Networks, or GNNs, to recognize features of satisfiability. To the best of our knowledge this has never been tried before. Our
more » ... findings are potentially profound. In a weakly-supervised setting, that is, without problem specific feature engineering, Graph Neural Networks can learn features of satisfiability.
arXiv:1702.03592v1 fatcat:gah2gjalzvb6bgz35qwp663u3i

Retrieval-guided Counterfactual Generation for QA [article]

Bhargavi Paranjape, Matthew Lamm, Ian Tenney
2022 arXiv   pre-print
In this work, we employ the QED formalism for explanations in question answering (Lamm et al., 2021) for query meaning representation.  ...  The results are fully labeled, and can be used directly to augment training data or filtered post-hoc for analysis using simple heuristics or meaning representations such as QED (Lamm et al., 2021) .  ... 
arXiv:2110.07596v2 fatcat:upw2ly4dhjeehb4rndukmuupuy

The Pragmatics of Indirect Commands in Collaborative Discourse [article]

Matthew Lamm, Mihail Eric
2017 arXiv   pre-print
Matthew Lamm * Stanford Linguistics Stanford NLP Group  ... 
arXiv:1705.03454v2 fatcat:dys5lmvhc5e2vbtxnxn2ald6le

Opportunities for Advanced Imaging Techniques in Pharmaceutical Formulation Development

Matthew S. Lamm
2019 Microscopy and Microanalysis  
The complexity of modern pharmaceutical formulation development requires increasingly sophisticated understanding of the role of composition and process on the ultimate performance of the drug product. Among the variety of analytical tools available to the pharmaceutical materials scientist, imaging plays a pivotal role in characterizing formulations precisely because of the richness of information contained within the image, especially when coupled with spectroscopic or other chemical
more » ... In a formulated tablet, for example, the ability to visualize the size and shape of the components within the formulation may provide useful information as to how the individual particles have been altered as result of milling or compaction processes. However, it is the understanding of the spatial distribution of each component relative to each other that can provide deeper insight as to how effectively the materials were blended and perhaps what physical or chemical changes may be occurring at the interface of two different types of particles such as an active ingredient (drug) and excipient. In this regard, spectroscopic chemical imaging is of particular value and is continuing to gain interest within the pharmaceutical field. The Merck Preformulation group has been evaluating chemical imaging techniques beyond traditional Raman microscopy that enable faster acquisition speeds but also provide good chemical specificity such as broadband coherent anti-stokes Raman spectroscopy (CARS) microscopy[1], as well as more recently a study evaluating the ability of stimulated Raman spectroscopy (SRS) imaging to visualize the dissolution process of drug crystals embedded in a controlled release implant [2] . In addition to fast Raman imaging techniques, infrared microscopy enabled by tuneable quantum cascade lasers (QCL) can provide similar but complementary spectroscopic information with rapid acquisition times. There are multiple commercial instruments now utilizing the recent advancements in QCL technology of which have been explored for use in pharmaceutical applications. In addition to chemical spectroscopic imaging techniques, X-ray computed tomography (XRCT) has become an increasingly important tool to understand the 3-dimensional structures of pharmaceutical formulations in a non-destructive quality control type of application where tablets are analysed for internal cracks, for example, or drug delivery devices such as inhalers are imaged to ensure appropriate dose filling. When paired with advanced image analysis and modelling/simulation, XRCT can be a very powerful tool to understand and predict the mechanism and rates of release of long acting implant formulations as both compositional and processing variables are explored [3] . For applications requiring high spatial resolution, scanning probe microscopy (SPM) can provide insight as to morphology and heterogeneity in multicomponent systems where traditional electron microscopy or chemical imaging techniques suffer from lack of contrast, spectral specificity or resolution. Within pharmaceutical formulation development, scanning probe techniques are readily applied to cases where the phase behaviour of drug, excipient, and polymers mixtures are examined especially when prepared using hot melt extrusion techniques. For these cases, alternate contact or tapping mode atomic force microscopy (AFM) with phase imaging can provide insight as to the extent of mixing and domain size of phase separated systems [4] . Furthermore, advanced scanning modes such as bimodal AFM imaging 1268
doi:10.1017/s1431927619007074 fatcat:7ez2kfqr65dj5pd2pwmykrl4ya

Measuring Attribution in Natural Language Generation Models [article]

Hannah Rashkin, Vitaly Nikolaev, Matthew Lamm, Michael Collins, Dipanjan Das, Slav Petrov, Gaurav Singh Tomar, Iulia Turc, David Reitter
2021 arXiv   pre-print
With recent improvements in natural language generation (NLG) models for various applications, it has become imperative to have the means to identify and evaluate whether NLG output is only sharing verifiable information about the external world. In this work, we present a new evaluation framework entitled Attributable to Identified Sources (AIS) for assessing the output of natural language generation models, when such output pertains to the external world. We first define AIS and introduce a
more » ... o-stage annotation pipeline for allowing annotators to appropriately evaluate model output according to AIS guidelines. We empirically validate this approach on three generation datasets (two in the conversational QA domain and one in summarization) via human evaluation studies that suggest that AIS could serve as a common framework for measuring whether model-generated statements are supported by underlying sources. We release guidelines for the human evaluation studies.
arXiv:2112.12870v1 fatcat:u63xkswswvbmpd5djbhoukfx4q

Textual Analogy Parsing: What's Shared and What's Compared among Analogous Facts [article]

Matthew Lamm, Arun Tejasvi Chaganty, Christopher D. Manning, Dan Jurafsky, Percy Liang
2018 arXiv   pre-print
As our SRL representation of quantitative facts, we employ the Quantitative Semantic Role Labeling (QSRL) framework we previously defined in Lamm et al. (2018) .  ...  semantic role structures in Figure 3 , but representing them separately in this way fails to capture the fact that Two analogous quantitative facts represented independently, using the QSRL schema (Lamm  ... 
arXiv:1809.02700v1 fatcat:qocei6hh7rg2zcdhoawberccpm

Characterizing Dense Nanosuspensions using Cryo-SEM

Nazia N. Khawaja, David J. Goldfarb, Sachin Mittal, Irina Kazakevich, Matthew Lamm
2018 Microscopy and Microanalysis  
Cryogenic-Scanning Electron Microscopy (cryo-SEM) is a technique which unlike standard SEM, allows the imaging of samples in their hydrated state. This technique was used to image dense nanosuspensions. Standard SEM images were captured previously, however, were questionable in value, because the sample needed to be dried prior to measurement. There were concerns that the particles could be altered upon drying, especially in the presence of surfactant. By using the cryo-SEM technique, this
more » ... g step was eliminated and the particles were able to be viewed in situ. The nanosuspension particle morphology, imaged with a Hitachi SU 5000 scanning electron microscope equipped with a Quorum PP3010T cryo unit, correlated with particle size measurements obtained with an orthogonal Dynamic Light Scattering (DLS) technique. It was also previously shown for these nanosuspensions that rheological measurements correlated with median particle size as measured by DLS. The DLS measurements for one nanosuspension formulation exhibited a median particle size growth and a corresponding increase in viscosity during accelerated stability studies. Two theories were hypothesized as to the cause for this phenomenon: aggregation of particles or Ostwald ripening. With the availability of cryo-SEM images, aggregation was ruled out. It is suspected that the morphology and size differences between these batches, observed by SEM and DLS, resulted in an increased viscosity ( Figure 1 ). In general, cryo-SEM was able to provide unique images that could not have been produced using traditional microscopy techniques, proving to be a useful tool for viewing nanoparticles in situ.
doi:10.1017/s1431927618005019 fatcat:cvt3vjnosndnfgrvdis6vmmgqe

QED: A Framework and Dataset for Explanations in Question Answering [article]

Matthew Lamm, Jennimaria Palomaki, Chris Alberti, Daniel Andor, Eunsol Choi, Livio Baldini Soares, Michael Collins
2020 arXiv   pre-print
., 2004; Lamm et al., 2018; Tandon et al., 2019) relations. Future uses of QED representations Our hope is that QED representations may be useful in a variety of extensions to extant QA systems.  ... 
arXiv:2009.06354v1 fatcat:mjoi4dao3rgzde5zw6rk63nleq

Public space in an age of austerity

Matthew Carmona, Gro Sandkjær Hanssen, Bettina Lamm, Katarina Nylund, Inger-Lise Saglie, Anne Tietjen
2019 Urban Design International  
Fig. 2 2 Triangeln, Möllevången: a new public space dating from 2010 in the centrally located dense working class district of Möllevången (Matthew Carmona) Fig. 3 Malmö Live, University Island: Fig.  ...  Carmona) Fig. 11 11 The rise in rough sleepers represents a direct and highly visible impact of austerity (Matthew Carmona)Fig. 12 Verdensparken (World park), Furuset, an upgrading of postwar green  ... 
doi:10.1057/s41289-019-00082-w fatcat:fijjpi5jo5dyjmwgtclsuzjeoi

Theories and methods in the scientific study of consciousness [chapter]

P. Christiaan Klink, Matthew W. Self, Victor A.F. Lamme, Pieter R. Roelfsema
2015 Advances in Consciousness Research  
, 2003 Lamme, , 2006 Lamme, , 2010 .  ...  , 2006 (Lamme, , 2010 .  ... 
doi:10.1075/aicr.92.02kli fatcat:57ch3bvqwveqxc4bia4n55tap4

Decontextualization: Making Sentences Stand-Alone

Eunsol Choi, Jennimaria Palomaki, Matthew Lamm, Tom Kwiatkowski, Dipanjan Das, Michael Collins
2021 Transactions of the Association for Computational Linguistics  
Models for question answering, dialogue agents, and summarization often interpret the meaning of a sentence in a rich context and use that meaning in a new context. Taking excerpts of text can be problematic, as key pieces may not be explicit in a local window. We isolate and define the problem of sentence decontextualization: taking a sentence together with its context and rewriting it to be interpretable out of context, while preserving its meaning. We describe an annotation procedure,
more » ... data on the Wikipedia corpus, and use the data to train models to automatically decontextualize sentences. We present preliminary studies that show the value of sentence decontextualization in a user-facing task, and as preprocessing for systems that perform document understanding. We argue that decontextualization is an important subtask in many downstream applications, and that the definitions and resources provided can benefit tasks that operate on sentences that occur in a richer context.
doi:10.1162/tacl_a_00377 fatcat:xq5kog3jlrcbncn43stej3lwzy

Ellipsis Resolution as Question Answering: An Evaluation [article]

Rahul Aralikatte, Matthew Lamm, Daniel Hardt, Anders Søgaard
2021 arXiv   pre-print
Most, if not all forms of ellipsis (e.g., so does Mary) are similar to reading comprehension questions (what does Mary do), in that in order to resolve them, we need to identify an appropriate text span in the preceding discourse. Following this observation, we present an alternative approach for English ellipsis resolution relying on architectures developed for question answering (QA). We present both single-task models, and joint models trained on auxiliary QA and coreference resolution
more » ... ts, clearly outperforming the current state of the art for Sluice Ellipsis (from 70.00 to 86.01 F1) and Verb Phrase Ellipsis (from 72.89 to 78.66 F1).
arXiv:1908.11141v3 fatcat:vmo6vvjh4baqlajmhfdgw6kghq

Stabilized imaging of immune surveillance in the mouse lung

Mark R Looney, Emily E Thornton, Debasish Sen, Wayne J Lamm, Robb W Glenny, Matthew F Krummel
2010 Nature Methods  
stabilized imaging of immune surveillance in the mouse lung Mark R Looney 1,4 , Emily E Thornton 2,4 , Debasish Sen 2 , Wayne J Lamm 3 , Robb W Glenny 3 & Matthew F Krummel 2 Compared to the sub-micrometer  ... 
doi:10.1038/nmeth.1543 pmid:21151136 pmcid:PMC3076005 fatcat:mxppr4q4jjad5a2f4ozafv2wee

Peptide−Silica Hybrid Networks: Biomimetic Control of Network Mechanical Behavior

Aysegul Altunbas, Nikhil Sharma, Matthew S. Lamm, Congqi Yan, Radhika P. Nagarkar, Joel P. Schneider, Darrin J. Pochan
2009 ACS Nano  
Self-assembly represents a robust and powerful paradigm for the bottom-up construction of nanostructures. Templated condensation of silica precursors on self-assembled nanoscale peptide fibrils with various surface functionalities can be used to mimic biosilicification. This template-defined approach toward biomineralization was utilized for the controlled fabrication of 3D hybrid nanostructures. The peptides MAX1 and MAX8 used herein form networks consisting of interconnected, self-assembled
more » ... ta-sheet fibrils. We report a study on the structure--property relationship of self-assembled peptide hydrogels where mineralization of individual fibrils through sol--gel chemistry was achieved. The nanostructure and consequent mechanical characteristics of these hybrid networks can be modulated by changing the stoichiometric parameters of the sol--gel process. The physical characterization of the hybrid networks via electron microscopy and small-angle scattering is detailed and correlated with changes in the network mechanical behavior. The resultant high fidelity templating process suggests that the peptide substrate can be used to template the coating of other functional inorganic materials.
doi:10.1021/nn901226h pmid:20028097 pmcid:PMC3488860 fatcat:u2zmaargp5gkhluu2hovoz33j4

Constraints for Time-Multiplexed Structured Light with a Hand-held Camera

Sepehr Ghavam, Matthew Post, Mohamed A. Naiel, Mark Lamm, Paul Fieguth
2021 Journal of Computational Vision and Imaging Systems  
Multi-frame structured light in projector-camera systems affords high-density and non-contact methods of 3D surface reconstruction. However, they have strict setup constraints which can become expensive and time-consuming. Here, we investigate the conditions under which a projective homography can be used to compensate for small perturbations in pose caused by a hand-held camera. We synthesize data using a pinhole camera model and use it to determine the average 2D reprojection error per point
more » ... orrespondence. This error map is grouped into regions with specified upper-bounds to classify which regions produce sufficiently minimal error to be considered feasible for a structured-light projector-camera system with a hand-held camera. Empirical results demonstrate that a sub-pixel reprojection accuracy is achievable with a feasible geometric constraints
doi:10.15353/jcvis.v6i1.3562 fatcat:y2hf62ymhvbjzef4grsxdevq2u
« Previous Showing results 1 — 15 out of 1,401 results