Nikhil Singh

Dartmouth College

Unclaimed · AI-structured profileOpenAlex

Dartmouth researcher linking representation geometry and neurosymbolic benchmarks across machine learning and NLP

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-13) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Insufficient data: this snapshot contains no publication-volume or authorship records beyond the two reviewed papers. The output axis would measure total papers, authorship positions, and production volume over time; no score is estimated.

Impact—

Insufficient data: no citation counts, h-index, or influence metrics are available. The impact axis would measure how widely the work is cited and taken up by other researchers; it is left unscored rather than guessed.

Focus0/100

The focus score is 0, and the raw field distribution explains it directly: recorded output splits exactly 50/50 between natural language processing / neurosymbolic AI and machine learning (representation learning). This is maximal spread across the two tracked fields — a balanced dual agenda rather than concentration in one area. The two 2026 papers mirror the split, one per field.

Collaboration46/100

Collaboration scores 46, with 9 distinct co-authors recorded. Frequent collaborators include Amartya Bhattacharya, Neeti Pokhriyal, Soroush Vosoughi, Chenming Shang, Yujin Tang, Jun Jie Ou Yang, Ruize Xu, and Adam Breuer. On the two 2026 papers he is last author once and middle author once, neither corresponding — a moderate, team-embedded collaboration pattern.

Activity—

Insufficient data: no publication-cadence or recency metrics are available beyond the two 2026-dated papers. The activity axis would measure publishing frequency and output recency over time; no level of current activity is inferred from this snapshot.

Reproducibility78/100

Reproducibility scores 78, the mean over 2 scored papers. The displacement-geometry paper is rated high reproducibility readiness; PRISM-BN is rated medium, with Evolve noting it is carefully built and honest while cautioning that reference Bayesian networks are LLM-generated and not externally validated.

Research summary

Nikhil Singh is a researcher at Dartmouth College whose recorded work sits evenly at the intersection of representation learning and natural language processing: the snapshot's field distribution splits exactly 50/50 between machine learning (representation learning) and NLP/neurosymbolic AI. Evolve reviews two 2026 arXiv papers from him. In 'Displacement Geometry Captures Platonic Shared Reality Across Models and Modalities' he is the last author; the paper argues that models share displacement directions rather than absolute positions, examined across 44 encoders, and thereby reframes the Platonic Representation Hypothesis debate. Evolve judges the argument compelling while flagging whitening and data overlap for scrutiny, and rates its reproducibility readiness high. The second paper, PRISM-BN, on which he is a middle author, delivers a carefully built, honest controlled corpus and benchmark for text-to-parameterized Bayesian network extraction; its stated limitation is that reference networks are LLM-generated and not externally validated, and its readiness is rated medium. Quantitatively, the snapshot records 9 distinct co-authors, a collaboration score of 46, a focus score of 0 arising from the even two-field split, and a reproducibility score of 78 averaged over 2 scored papers. Output, impact, and activity metrics are not yet available, so career-stage or citation-based judgments should await fuller data.

Collaboration pattern

The snapshot records 9 distinct co-authors and a collaboration score of 46. Frequent co-authors on record include Amartya Bhattacharya, Neeti Pokhriyal, Soroush Vosoughi, Chenming Shang, Yujin Tang, Jun Jie Ou Yang, Ruize Xu, and Adam Breuer. On the two 2026 papers he appears once as last author and once as a middle author, neither corresponding — suggesting contribution within multi-author teams rather than solo-led projects.

Research keywords

Platonic Representation Hypothesisrepresentational alignmentdisplacement vectorsProcrustes alignmentconcept geometryrelational distillationBayesian networkstext-to-BN extractionbenchmark datasetprobabilistic graphical modelsCPD elicitationneurosymbolic AI

Highlights

Reframing the Platonic Representation Hypothesis with displacement geometry
As last author on this 2026 arXiv paper, he contributes to the argument that models share displacement directions rather than absolute positions, examined across 44 encoders. Evolve finds the case compelling while flagging whitening and data overlap for scrutiny, and rates reproducibility readiness high.
PRISM-BN: a controlled benchmark for text-to-parameterized Bayesian network extraction
A middle-author contribution to a carefully built, honest controlled corpus and benchmark for extracting parameterized Bayesian networks from text. Evolve's key caveat is that the reference Bayesian networks are LLM-generated and not externally validated. Reproducibility readiness is rated medium.
A balanced two-front agenda: concept geometry and structured extraction
The snapshot's field distribution splits exactly 50/50 between machine learning (representation learning) and NLP/neurosymbolic AI, and each 2026 paper advances one front — aligning model representations versus benchmarking structured extraction. The recorded pattern includes 9 distinct co-authors and a collaboration score of 46.

Timeline

2026Last-author paper on displacement geometry across 44 encoders appears on arXiv; Evolve rates its reproducibility readiness high.
2026Middle-author contribution to PRISM-BN, a controlled corpus and benchmark for text-to-parameterized Bayesian network extraction.
2026Both reviewed works are dated 2026, spanning the profile's two fields — representation learning and NLP/neurosymbolic AI — within a single year.

Papers in the Evolve catalog

Frequent co-authors

Amartya Bhattacharya×1 Neeti Pokhriyal×1 Soroush Vosoughi×1 Nikhil Singh×1 Chenming Shang×1 Yujin Tang×1 Jun Jie Ou Yang×1 Ruize Xu×1 Adam Breuer×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometric scores come only from this platform's bibliometric snapshot; output, impact, and activity axes lack sufficient data and are unscored.
No ORCID or country is provided, and a same-name entry appears in the co-author list, so this profile is name-resolved from public bibliographic data for 'Nikhil Singh' at Dartmouth College.
Reproducibility (78) is a mean over only 2 Evolve-reviewed papers; the catalog may not cover this researcher's full publication record.
Both reviewed works are 2026 arXiv preprints; peer-review status and final venues are not recorded.