Papers › NBDT: Neural-Backed Decision Trees

NBDT: Neural-Backed Decision Trees

1 Apr 2020arXiv:2004.00221archive 2025-07-28

Alvin Wan, Lisa Dunlap, Daniel Ho, Jihan Yin, Scott Lee, Henry Jin, Suzanne Petryk, Sarah Adel Bargal, Joseph E. Gonzalez

Machine learning applications such as finance and medicine demand accurate and justifiable predictions, barring most deep learning methods from use. In response, previous work combines decision trees with deep learning, yielding models that (1) sacrifice interpretability for accuracy or (2) sacrifice accuracy for interpretability. We forgo this dilemma by jointly improving accuracy and interpretability using Neural-Backed Decision Trees (NBDTs). NBDTs replace a neural network's final linear layer with a differentiable sequence of decisions and a surrogate loss. This forces the model to learn high-level concepts and lessens reliance on highly-uncertain decisions, yielding (1) accuracy: NBDTs match or outperform modern neural networks on CIFAR, ImageNet and better generalize to unseen classes by up to 16%. Furthermore, our surrogate loss improves the original model's accuracy by up to 2%. NBDTs also afford (2) interpretability: improving human trustby clearly identifying model mistakes and assisting in dataset debugging. Code and pretrained NBDTs are at https://github.com/alvinwan/neural-backed-decision-trees.

PaperPDFCodeCode Syntology ranMethod as code

In Syntology View this paper on Syntology, its page in Syntology's graph. That page adds which of its functions are also present in other papers' code Syntology holds, with dates, and its dated run record as JSON and a README badge. It also lists the repositories linked to the paper, the abstract and the calls for agents.

Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, on Syntology's MCP service (how to connect):

Code

Syntology 29 code samples harvested from 2 repositories linked to this paper: 18 ran (of which 2 constructed an object rather than computing a result; 4 with no instrument failure: 1 honoured, 0 violated, 3 with no contract checked; 14 where Syntology's instrument failed) · 11 unverified. Where Syntology's instrument failed: 14 where our contract draft was wrong; failures of Syntology's instrument, not of the code.

By repository: official repository: 6 samples from 1 repository, 1 ran (0 with no instrument failure; 1 where Syntology's instrument failed); community (archive-listed): 23 samples from 1 repository, 17 ran (4 with no instrument failure; 13 where Syntology's instrument failed). The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

No sample is linked to this paper from 1 of its repositories that Syntology harvested for other papers:

Levels written on 2026-09-22 (UTC): the dates Syntology's graph wrote each sample's verification level, not when the code executed.

alvinwan/neural-backed-decision-trees officialmentioned in papermentioned on GitHubpytorchSyntology: 6 samples, 1 ran (1 instrument-failed) · 5 unverified report
XAVILLA/nbdt mentioned on GitHubpytorchSyntology: 23 samples, 17 ran (13 instrument-failed) · 6 unverified report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

A paper named beside a repository with no sample linked to this paper is shown with that paper's own run record, not this paper's: “ran” means executed on a synthesized input, not that the code is correct, and “where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Papers are listed in arXiv-id order, at most three per repository.

Code Syntology ran Syntology

29 samples harvested: 18 ran (of which 2 constructed an object rather than computing a result; 4 with no instrument failure: 1 honoured, 0 violated, 3 with no contract checked; 14 where Syntology's instrument failed) · 11 unverified. “Where Syntology's instrument failed” counts failures of Syntology's instrument, not of the code. Syntology's record for this page has not changed since , the first build that kept a record date for it; when this build read Syntology's graph is in the build record. That date is not when the samples ran.

1ran · honoured contract
14ran · our draft was wrong
3ran
11unverified

Licence: 0 of the 29 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from 2 repositories linked to this paper, official or community; the list below is grouped by repository, each group headed by its name and what it is to the paper. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. A line beginning “fixture:” under a sample that ran gives the arguments Syntology's fixture called the sample with, each as name:type/rank/element kind (? where none was recorded), and the type of what the fixture's run returned, exactly as the harness recorded them; the arguments and the returned type were recorded by two separate runs of the same fixture. No output value is shown, and a returned type says nothing about whether the output was right. “Level recorded” gives the UTC date when Syntology's graph wrote the sample's verification level, not when the code executed; where the level was written before Syntology's graph kept that date, the row says “level recorded, date not recorded”. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself. “Our draft was wrong” and “the contract check could not run” are failures of Syntology's instrument, not of the code: the sample ran on Syntology's synthesized input, and then the drafted contract did not describe it, or the check itself raised an error (RAISES), lacked a dependency (DEP_MISSING) or ran out of time (TIMEOUT); hover the label for which. A metamorphic tier, where shown, is evidence of the output's shape, not that the sample implements the method.

Each sample ends with its code_sha256. A sample's code_sha256 is Syntology's identity for that exact code. It is printed in full (16 hexadecimal characters for every sample in this build), and get_code(code_sha256="…") takes it exactly as printed. An agent calls it on Syntology's MCP service (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples are grouped by repository: a repository marked official first, then by how directly the repository is tied to the paper, then in the archive's order; never by how many ran. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of six reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar); this paper's copy was not located, so its licence is not recorded; or this paper's copy is cleared, but the stored copy of the same code that Syntology serves is not. Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

A function or class name links to its page when the name occurs in the harvested code of at least 5 papers; the name is a string, not an identity.

alvinwan/neural-backed-decision-trees official repository 6 samples, 1 ran (1 instrument-failed) · 5 unverified
dataset_to_dummy_classes nbdt/model.py ran · our draft was wrong MIT (permissive) · a32351468906ce8a · report fixture: signature dataset:str/0/str · returned list · level recorded 2026-09-22
EmbeddedDecisionRules nbdt/model.py unverified MIT (permissive) · 55ce848c52ee5cb8 · report
HardEmbeddedDecisionRules nbdt/model.py unverified MIT (permissive) · be59a6ad0e2ddec0 · report
NBDT nbdt/model.py unverified MIT (permissive) · 899f401f38f1fd83 · report
Tree in 10 papers nbdt/model.py unverified MIT (permissive) · fbc1d6876bb0df31 · report
load_state_dict_from_key nbdt/model.py unverified MIT (permissive) · 14f8f9f850c7e428 · report
XAVILLA/nbdt community (archive-listed) 23 samples, 17 ran (13 instrument-failed) · 6 unverified
Colors nbdt/model.py ran MIT (permissive) · 1020d5e79c4766de · report fixture: returned Colors · level recorded 2026-09-22
FakeSynset nbdt/model.py ran MIT (permissive) · 52cf40b8aae2967f · report fixture: returned FakeSynset · level recorded 2026-09-22
coerce_state_dict nbdt/model.py ran · our draft was wrong MIT (permissive) · 303cf42ae5edc004 · report fixture: signature state_dict:dict/0/?, reference_state_dict:dict/0/? · returned dict · level recorded 2026-09-22
dataset_to_default_path_graph nbdt/model.py ran · our draft was wrong MIT (permissive) · 23b18dbf212e73e9 · report fixture: returned str · level recorded 2026-09-22
dataset_to_default_path_wnids nbdt/model.py ran · our draft was wrong MIT (permissive) · e081d64f5cad1cd9 · report fixture: returned str · level recorded 2026-09-22
dataset_to_dummy_classes nbdt/model.py ran · our draft was wrong MIT (permissive) · 4625fb450d4c7b8c · report fixture: signature dataset:str/0/str · returned list · level recorded 2026-09-22
fwd nbdt/model.py ran MIT (permissive) · 8497ac1255b3b1a5 · report fixture: returned PosixPath · level recorded 2026-09-22
get_leaf_to_path nbdt/model.py ran · our draft was wrong MIT (permissive) · 642244caa777143c · report fixture: signature G:DiGraph/0/? · returned dict · level recorded 2026-09-22
get_leaves nbdt/model.py ran · our draft was wrong MIT (permissive) · 6cbac319834ba8ae · report fixture: signature G:DiGraph/0/?, root:NoneType/0/? · returned generator · level recorded 2026-09-22
get_roots nbdt/model.py ran · our draft was wrong MIT (permissive) · e31c7cca6d6133a2 · report fixture: signature G:DiGraph/0/? · returned generator · level recorded 2026-09-22
get_wnids nbdt/model.py ran · our draft was wrong MIT (permissive) · e270d7f9006e5c50 · report fixture: signature path_wnids:str/0/str · returned list · level recorded 2026-09-22
hierarchy_to_path_graph nbdt/model.py ran · our draft was wrong MIT (permissive) · 8c79e86eb84f0948 · report fixture: returned str · level recorded 2026-09-22
is_leaf in 11 papers nbdt/model.py ran · our draft was wrong MIT (permissive) · 626c774cdb0ecdea · report fixture: signature G:DiGraph/0/?, node:int/0/int · returned bool · level recorded 2026-09-22
read_graph in 11 papers nbdt/model.py ran · honoured contract MIT (permissive) · 7ef704b1f90df1c8 · report fixture: signature path:str/0/str · returned Graph · level recorded 2026-09-22
synset_to_name nbdt/model.py ran · our draft was wrong MIT (permissive) · 8d5c8808f6873339 · report fixture: signature synset:MockSynset/0/? · returned str · level recorded 2026-09-22
wnid_to_name nbdt/model.py ran · our draft was wrong fingerprinted MIT (permissive) · 71ff6b40cffae955 · report fixture: signature wnid:str/0/str · returned str · level recorded 2026-09-22
wnid_to_synset nbdt/model.py ran · our draft was wrong MIT (permissive) · 1743b41d3adb91a6 · report fixture: signature wnid:str/0/str · returned FakeSynset · level recorded 2026-09-22
EmbeddedDecisionRules nbdt/model.py unverified MIT (permissive) · 95c098426e5a427f · report
HardEmbeddedDecisionRules nbdt/model.py unverified MIT (permissive) · e9189fbf051e129d · report
NBDT nbdt/model.py unverified MIT (permissive) · 13e0602b87690b50 · report
Node in 32 papers nbdt/model.py unverified MIT (permissive) · 20d642b846aa125d · report
Tree in 10 papers nbdt/model.py unverified MIT (permissive) · 4553e20f251d19cb · report
load_state_dict_from_key nbdt/model.py unverified MIT (permissive) · e33501c9d70eab4b · report

Method as code Syntology

What Syntology's graph records as this paper's claimed method, lifted into code: each assembly names its bricks (functions and classes), each with its own role and its own verification level. An assembly is not a check that the code implements the method correctly or reproduces the paper's results. Levels are each brick's own verification level in Syntology's graph (level 0: no recorded run), listed brick by brick and never combined into one level; “ran” means the brick executed on a synthesized input, not that its output is correct, and the instrument figure counts failures of Syntology's instrument, not of the code. Where a model read the code against the paper's claim, each assembly's label verdict is one model's judgement, kept in this page's JSON twin; it is not calibrated, so no card prints it. Siblings are assemblies lifted from this paper in the same batch for the same method: alternative readings kept side by side; none replaces or supersedes another.

2 assemblies on this paper: 2 harvested, 0 generated, listed in 1 groups (siblings lifted in one batch form one group).

2 sibling assemblies for Neural-Backed Decision Trees (NBDTs) (harvested), lifted in the same batch on 2026-09-09: alternative readings kept side by side; none replaces or supersedes another.

harvested Assembly for Neural-Backed Decision Trees (NBDTs) (graph method: Neural-Backed Decision Trees (NBDTs)). 23 bricks; individually on a fixture: 17 ran (of which 2 constructed an object rather than computing a result; 4 with no instrument failure: 1 honoured, 0 violated, 3 with no contract checked; 13 where Syntology's instrument failed) · 6 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. NBDT entry · level 0 · unverified · 13e0602b87690b50
  2. HardEmbeddedDecisionRules model · level 0 · unverified · e9189fbf051e129d
  3. Tree helper · level 0 · unverified · 4553e20f251d19cb
  4. coerce_state_dict helper · level 1 · ran · our draft was wrong · 303cf42ae5edc004
  5. load_state_dict_from_key helper · level 0 · unverified · e33501c9d70eab4b
  6. EmbeddedDecisionRules helper · level 0 · unverified · 95c098426e5a427f
  7. Node helper · level 0 · unverified · 20d642b846aa125d
  8. dataset_to_default_path_graph helper · level 1 · ran · our draft was wrong · 23b18dbf212e73e9
  9. dataset_to_default_path_wnids helper · level 1 · ran · our draft was wrong · e081d64f5cad1cd9
  10. dataset_to_dummy_classes helper · level 1 · ran · our draft was wrong · 4625fb450d4c7b8c
  11. get_leaf_to_path helper · level 1 · ran · our draft was wrong · 642244caa777143c
  12. get_wnids helper · level 1 · ran · our draft was wrong · e270d7f9006e5c50
  13. hierarchy_to_path_graph helper · level 1 · ran · our draft was wrong · 8c79e86eb84f0948
  14. read_graph helper · level 1 · ran · honoured contract · 7ef704b1f90df1c8
  15. Colors helper · level 1 · ran · 1020d5e79c4766de
  16. get_leaves helper · level 1 · ran · our draft was wrong · 6cbac319834ba8ae
  17. wnid_to_name helper · level 1 · ran · our draft was wrong · 71ff6b40cffae955
  18. wnid_to_synset helper · level 1 · ran · our draft was wrong · 1743b41d3adb91a6
  19. fwd helper · level 1 · ran · 8497ac1255b3b1a5
  20. FakeSynset helper · level 1 · ran · 52cf40b8aae2967f
  21. get_roots helper · level 1 · ran · our draft was wrong · e31c7cca6d6133a2
  22. is_leaf helper · level 1 · ran · our draft was wrong · 626c774cdb0ecdea
  23. synset_to_name helper · level 1 · ran · our draft was wrong · 8d5c8808f6873339

harvested Assembly for Neural-Backed Decision Trees (NBDTs) (graph method: Neural-Backed Decision Trees (NBDTs)). 23 bricks; individually on a fixture: 17 ran (of which 2 constructed an object rather than computing a result; 4 with no instrument failure: 1 honoured, 0 violated, 3 with no contract checked; 13 where Syntology's instrument failed) · 6 unverified. The assembly has not been run end-to-end.

one of 2 siblings

  1. NBDT entry · level 0 · unverified · 899f401f38f1fd83
  2. HardEmbeddedDecisionRules model · level 0 · unverified · be59a6ad0e2ddec0
  3. Tree helper · level 0 · unverified · fbc1d6876bb0df31
  4. coerce_state_dict helper · level 1 · ran · our draft was wrong · 303cf42ae5edc004
  5. load_state_dict_from_key helper · level 0 · unverified · 14f8f9f850c7e428
  6. EmbeddedDecisionRules model · level 0 · unverified · 55ce848c52ee5cb8
  7. Node helper · level 0 · unverified · 20d642b846aa125d
  8. dataset_to_default_path_graph helper · level 1 · ran · our draft was wrong · 23b18dbf212e73e9
  9. dataset_to_default_path_wnids helper · level 1 · ran · our draft was wrong · e081d64f5cad1cd9
  10. dataset_to_dummy_classes helper · level 1 · ran · our draft was wrong · a32351468906ce8a
  11. get_leaf_to_path helper · level 1 · ran · our draft was wrong · 642244caa777143c
  12. get_wnids helper · level 1 · ran · our draft was wrong · e270d7f9006e5c50
  13. hierarchy_to_path_graph helper · level 1 · ran · our draft was wrong · 8c79e86eb84f0948
  14. read_graph helper · level 1 · ran · honoured contract · 7ef704b1f90df1c8
  15. Colors helper · level 1 · ran · 1020d5e79c4766de
  16. get_leaves helper · level 1 · ran · our draft was wrong · 6cbac319834ba8ae
  17. wnid_to_name helper · level 1 · ran · our draft was wrong · 71ff6b40cffae955
  18. wnid_to_synset helper · level 1 · ran · our draft was wrong · 1743b41d3adb91a6
  19. fwd helper · level 1 · ran · 8497ac1255b3b1a5
  20. FakeSynset helper · level 1 · ran · 52cf40b8aae2967f
  21. get_roots helper · level 1 · ran · our draft was wrong · e31c7cca6d6133a2
  22. is_leaf helper · level 1 · ran · our draft was wrong · 626c774cdb0ecdea
  23. synset_to_name helper · level 1 · ran · our draft was wrong · 8d5c8808f6873339

Each brick ends with its code_sha256, Syntology's identity for that exact code; click one to copy the get_code call that returns it. For agents, on Syntology's MCP service: assembly(arxiv_id="2004.00221") returns these assemblies with each brick's role and wiring (how to connect).

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

1x1 ConvolutionAverage PoolingBatch NormalizationConvolutionDense ConnectionsDepthwise ConvolutionDepthwise Separable ConvolutionDropoutEfficientNetInterpretabilityInverted Residual BlockLinear LayerPointwise ConvolutionRMSPropReLUSigmoid ActivationSqueeze-and-Excitation Block

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections