-
Progressive Inference: Explaining Decoder-Only Sequence Classification Models Using Intermediate Predictions
Authors:
Sanjay Kariyappa,
Freddy Lécué,
Saumitra Mishra,
Christopher Pond,
Daniele Magazzeni,
Manuela Veloso
Abstract:
This paper proposes Progressive Inference - a framework to compute input attributions to explain the predictions of decoder-only sequence classification models. Our work is based on the insight that the classification head of a decoder-only Transformer model can be used to make intermediate predictions by evaluating them at different points in the input sequence. Due to the causal attention mechan…
▽ More
This paper proposes Progressive Inference - a framework to compute input attributions to explain the predictions of decoder-only sequence classification models. Our work is based on the insight that the classification head of a decoder-only Transformer model can be used to make intermediate predictions by evaluating them at different points in the input sequence. Due to the causal attention mechanism, these intermediate predictions only depend on the tokens seen before the inference point, allowing us to obtain the model's prediction on a masked input sub-sequence, with negligible computational overheads. We develop two methods to provide sub-sequence level attributions using this insight. First, we propose Single Pass-Progressive Inference (SP-PI), which computes attributions by taking the difference between consecutive intermediate predictions. Second, we exploit a connection with Kernel SHAP to develop Multi Pass-Progressive Inference (MP-PI). MP-PI uses intermediate predictions from multiple masked versions of the input to compute higher quality attributions. Our studies on a diverse set of models trained on text classification tasks show that SP-PI and MP-PI provide significantly better attributions compared to prior work.
△ Less
Submitted 3 June, 2024;
originally announced June 2024.
-
The Dark Energy Survey Data Release 2
Authors:
DES Collaboration,
T. M. C. Abbott,
M. Adamow,
M. Aguena,
S. Allam,
A. Amon,
J. Annis,
S. Avila,
D. Bacon,
M. Banerji,
K. Bechtol,
M. R. Becker,
G. M. Bernstein,
E. Bertin,
S. Bhargava,
S. L. Bridle,
D. Brooks,
D. L. Burke,
A. Carnero Rosell,
M. Carrasco Kind,
J. Carretero,
F. J. Castander,
R. Cawthon,
C. Chang,
A. Choi
, et al. (110 additional authors not shown)
Abstract:
We present the second public data release of the Dark Energy Survey, DES DR2, based on optical/near-infrared imaging by the Dark Energy Camera mounted on the 4-m Blanco telescope at Cerro Tololo Inter-American Observatory in Chile. DES DR2 consists of reduced single-epoch and coadded images, a source catalog derived from coadded images, and associated data products assembled from 6 years of DES sc…
▽ More
We present the second public data release of the Dark Energy Survey, DES DR2, based on optical/near-infrared imaging by the Dark Energy Camera mounted on the 4-m Blanco telescope at Cerro Tololo Inter-American Observatory in Chile. DES DR2 consists of reduced single-epoch and coadded images, a source catalog derived from coadded images, and associated data products assembled from 6 years of DES science operations. This release includes data from the DES wide-area survey covering ~5000 deg2 of the southern Galactic cap in five broad photometric bands, grizY. DES DR2 has a median delivered point-spread function full-width at half maximum of g= 1.11, r= 0.95, i= 0.88, z= 0.83, and Y= 0.90 arcsec photometric uniformity with a standard deviation of < 3 mmag with respect to Gaia DR2 G-band, a photometric accuracy of ~10 mmag, and a median internal astrometric precision of ~27 mas. The median coadded catalog depth for a 1.95 arcsec diameter aperture at S/N= 10 is g= 24.7, r= 24.4, i= 23.8, z= 23.1 and Y= 21.7 mag. DES DR2 includes ~691 million distinct astronomical objects detected in 10,169 coadded image tiles of size 0.534 deg2 produced from 76,217 single-epoch images. After a basic quality selection, benchmark galaxy and stellar samples contain 543 million and 145 million objects, respectively. These data are accessible through several interfaces, including interactive image visualization tools, web-based query clients, image cutout servers and Jupyter notebooks. DES DR2 constitutes the largest photometric data set to date at the achieved depth and photometric precision.
△ Less
Submitted 6 September, 2021; v1 submitted 14 January, 2021;
originally announced January 2021.
-
Dark Energy Survey Year 3 Results: Photometric Data Set for Cosmology
Authors:
I. Sevilla-Noarbe,
K. Bechtol,
M. Carrasco Kind,
A. Carnero Rosell,
M. R. Becker,
A. Drlica-Wagner,
R. A. Gruendl,
E. S. Rykoff,
E. Sheldon,
B. Yanny,
A. Alarcon,
S. Allam,
A. Amon,
A. Benoit-Lévy,
G. M. Bernstein,
E. Bertin,
D. L. Burke,
J. Carretero,
A. Choi,
H. T. Diehl,
S. Everett,
B. Flaugher,
E. Gaztanaga,
J. Gschwend,
I. Harrison
, et al. (89 additional authors not shown)
Abstract:
We describe the Dark Energy Survey (DES) photometric data set assembled from the first three years of science operations to support DES Year 3 cosmology analyses, and provide usage notes aimed at the broad astrophysics community. Y3 Gold improves on previous releases from DES, Y1 Gold and Data Release 1 (DES DR1), presenting an expanded and curated data set that incorporates algorithmic developmen…
▽ More
We describe the Dark Energy Survey (DES) photometric data set assembled from the first three years of science operations to support DES Year 3 cosmology analyses, and provide usage notes aimed at the broad astrophysics community. Y3 Gold improves on previous releases from DES, Y1 Gold and Data Release 1 (DES DR1), presenting an expanded and curated data set that incorporates algorithmic developments in image detrending and processing, photometric calibration, and object classification. Y3 Gold comprises nearly 5000 square degrees of grizY imaging in the south Galactic cap, including nearly 390 million objects, with depth reaching S/N ~ 10 for extended objects up to $i_{AB}\sim 23.0$, and top-of-the-atmosphere photometric uniformity $< 3$ mmag. Compared to DR1, photometric residuals with respect to Gaia are reduced by $50\%$, and per-object chromatic corrections are introduced. Y3 Gold augments DES DR1 with simultaneous fits to multi-epoch photometry for more robust galaxy color measurements and corresponding photometric redshift estimates. Y3 Gold features improved morphological star-galaxy classification with efficiency $>98\%$ and purity $>99\%$ for galaxies with $19 < i_{AB} < 22.5$. Additionally, it includes per-object quality information, and accompanying maps of the footprint coverage, masked regions, imaging depth, survey conditions, and astrophysical foregrounds that are used to select the cosmology analysis samples. This paper will be complemented by online resources.
△ Less
Submitted 27 May, 2021; v1 submitted 6 November, 2020;
originally announced November 2020.
-
The Dark Energy Survey Data Release 1
Authors:
T. M. C. Abbott,
F. B. Abdalla,
S. Allam,
A. Amara,
J. Annis,
J. Asorey,
S. Avila,
O. Ballester,
M. Banerji,
W. Barkhouse,
L. Baruah,
M. Baumer,
K. Bechtol,
M . R. Becker,
A. Benoit-Lévy,
G. M. Bernstein,
E. Bertin,
J. Blazek,
S. Bocquet,
D. Brooks,
D. Brout,
E. Buckley-Geer,
D. L. Burke,
V. Busti,
R. Campisano
, et al. (177 additional authors not shown)
Abstract:
We describe the first public data release of the Dark Energy Survey, DES DR1, consisting of reduced single epoch images, coadded images, coadded source catalogs, and associated products and services assembled over the first three years of DES science operations. DES DR1 is based on optical/near-infrared imaging from 345 distinct nights (August 2013 to February 2016) by the Dark Energy Camera mount…
▽ More
We describe the first public data release of the Dark Energy Survey, DES DR1, consisting of reduced single epoch images, coadded images, coadded source catalogs, and associated products and services assembled over the first three years of DES science operations. DES DR1 is based on optical/near-infrared imaging from 345 distinct nights (August 2013 to February 2016) by the Dark Energy Camera mounted on the 4-m Blanco telescope at Cerro Tololo Inter-American Observatory in Chile. We release data from the DES wide-area survey covering ~5,000 sq. deg. of the southern Galactic cap in five broad photometric bands, grizY. DES DR1 has a median delivered point-spread function of g = 1.12, r = 0.96, i = 0.88, z = 0.84, and Y = 0.90 arcsec FWHM, a photometric precision of < 1% in all bands, and an astrometric precision of 151 mas. The median coadded catalog depth for a 1.95" diameter aperture at S/N = 10 is g = 24.33, r = 24.08, i = 23.44, z = 22.69, and Y = 21.44 mag. DES DR1 includes nearly 400M distinct astronomical objects detected in ~10,000 coadd tiles of size 0.534 sq. deg. produced from ~39,000 individual exposures. Benchmark galaxy and stellar samples contain ~310M and ~ 80M objects, respectively, following a basic object quality selection. These data are accessible through a range of interfaces, including query web clients, image cutout servers, jupyter notebooks, and an interactive coadd image visualization tool. DES DR1 constitutes the largest photometric data set to date at the achieved depth and photometric precision.
△ Less
Submitted 23 April, 2019; v1 submitted 9 January, 2018;
originally announced January 2018.
-
The Dark Energy Survey Image Processing Pipeline
Authors:
E. Morganson,
R. A. Gruendl,
F. Menanteau,
M. Carrasco Kind,
Y. -C. Chen,
G. Daues,
A. Drlica-Wagner,
D. N. Friedel,
M. Gower,
M. W. G. Johnson,
M. D. Johnson,
R. Kessler,
F. Paz-Chinchón,
D. Petravick,
C. Pond,
B. Yanny,
S. Allam,
R. Armstrong,
W. Barkhouse,
K. Bechtol,
A. Benoit-Lévy,
G. M. Bernstein,
E. Bertin,
E. Buckley-Geer,
R. Covarrubias
, et al. (18 additional authors not shown)
Abstract:
The Dark Energy Survey (DES) is a five-year optical imaging campaign with the goal of understanding the origin of cosmic acceleration. DES performs a 5000 square degree survey of the southern sky in five optical bands (g,r,i,z,Y) to a depth of ~24th magnitude. Contemporaneously, DES performs a deep, time-domain survey in four optical bands (g,r,i,z) over 27 square degrees. DES exposures are proces…
▽ More
The Dark Energy Survey (DES) is a five-year optical imaging campaign with the goal of understanding the origin of cosmic acceleration. DES performs a 5000 square degree survey of the southern sky in five optical bands (g,r,i,z,Y) to a depth of ~24th magnitude. Contemporaneously, DES performs a deep, time-domain survey in four optical bands (g,r,i,z) over 27 square degrees. DES exposures are processed nightly with an evolving data reduction pipeline and evaluated for image quality to determine if they need to be retaken. Difference imaging and transient source detection are also performed in the time domain component nightly. On a bi-annual basis, DES exposures are reprocessed with a refined pipeline and coadded to maximize imaging depth. Here we describe the DES image processing pipeline in support of DES science, as a reference for users of archival DES data, and as a guide for future astronomical surveys.
△ Less
Submitted 9 January, 2018;
originally announced January 2018.
-
Metal-insulator transition and electroresistance in lanthanum/calcium manganites La_<1-x>Ca_<x>MnO_<3> (x = 0-0.5) from voltage-current-temperature surfaces
Authors:
J C Knott,
D C Pond,
R A Lewis
Abstract:
Of the perovskites, ABX_<3>, a subset of special interest is the family in which the A site is occupied by a lanthanide ion, the B site by a rare earth and X is oxygen, as such materials often exhibit a large change in electrical resistance in a magnetic field, a phenomenon known as "colossal" magnetoresistance (MR). Two additional phenomena in this family have also drawn attention: the metal-in…
▽ More
Of the perovskites, ABX_<3>, a subset of special interest is the family in which the A site is occupied by a lanthanide ion, the B site by a rare earth and X is oxygen, as such materials often exhibit a large change in electrical resistance in a magnetic field, a phenomenon known as "colossal" magnetoresistance (MR). Two additional phenomena in this family have also drawn attention: the metal-insulator transition (MIT) and electroresistance (ER). The MIT is revealed by measuring resistance as a function of temperature, and observing a change in the sign of the gradient. ER - the dependence of the resistance on applied current - is revealed by measuring resistance as a function of applied current. Up until now, the phenomena of MIT and ER have been treated separately. Here we report simultaneous observation of the MIT and ER in the lanthanum/calcium manganites. We accomplish this by measuring voltage-current curves over a wide temperature range (10-300 K) allowing us to build up an experimental voltage surface over current-temperature axes. These data directly lead to resistance surfaces. This approach provides additional insight into the phenomena of electrical transport in the lanthanum/calcium manganites, in particular the close connection of the maximum ER to the occurrence of the MIT in those cases of a paramagnetic insulator (PMI) to ferromagnetic metal (FMM) transition.
△ Less
Submitted 15 January, 2009;
originally announced January 2009.
