Publications (74)
Stochastic Negative Mining for Learning with Large Output Spaces
Sashank J. Reddi, Satyen Kale, Felix Yu +3
We consider the problem of retrieving the most relevant labels for a given input when the size of the output space is very large. Retrieval methods are modeled as set-valued classi…
Doppler Shifted Unruh Radiation: The Non-Relativistic Spectrum
Jeff S. Lee, Gerald B. Cleaver, Felix Yu
In this note, the non-relativistic Doppler spectrum of the Unruh radiation from an accelerating mass is determined. Additionally, the scattering of thermal bath photons off an acce…
Probing new gauge symmetries via exotic decays
Lisa Michaels, Felix Yu
New gauge theories involving Standard Model (SM) fermions typically require additional electroweak fermions for anomaly cancellation. We study the non-decoupling properties…
Modifying Memories in Transformer Models
Chen Zhu, Ankit Singh Rawat, Manzil Zaheer +4
Large Transformer models have achieved impressive performance in many natural language tasks. In particular, Transformer based language models have been shown to have great capabil…
Compatibility of theta13 and the Type I Seesaw Model with A4 Symmetry
Mu-Chun Chen, Jinrui Huang, Jon-Michael O'Bryan +2
We derive formulae for neutrino masses and mixing angles in a type I seesaw framework with an underlying A4 flavor symmetry. In particular, the Majorana neutrino mass matrix includ…
The Lazy Neuron Phenomenon: On Emergence of Activation Sparsity in Transformers
Zonglin Li, Chong You, Srinadh Bhojanapalli +8
This paper studies the curious phenomenon for machine learning models with Transformer architectures that their activation maps are sparse. By activation map we refer to the interm…
Phenomenology of Enhanced Light Quark Yukawa Couplings and the Charge Asymmetry
Felix Yu
I propose the measurement of the charge asymmetry as a consistency test for the Standard Model (SM) Higgs, which is sensitive to enhanced Yukawa couplings of the first an…
ReST meets ReAct: Self-Improvement for Multi-Step Reasoning LLM Agent
Renat Aksitov, Sobhan Miryoosefi, Zonglin Li +10
Answering complex natural language questions often necessitates multi-step reasoning and integrating external information. Several systems have combined knowledge retrieval with a…
cpSGD: Communication-efficient and differentially-private distributed SGD
Naman Agarwal, Ananda Theertha Suresh, Felix Yu +2
Distributed stochastic gradient descent is an important subroutine in distributed learning. A setting of particular interest is when the clients are mobile devices, where two impor…
Collider constraints and new tests of color octet vectors
Malte Buschmann, Felix Yu
We analyze the collider sensitivity for new colored resonances in , , and final states. While searches in the single production channel are model-dependen…
R-symmetry Matching In SUSY Breaking Models
Jessica Goodman, Masahiro Ibe, Yuri Shirman +1
Low energy descriptions of metastable supersymmetry breaking models often possess an accidental R-symmetry. Viable phenomenological applications of this class of models require R-s…
The Coannihilation Codex
Michael J. Baker, Joachim Brod, Sonia El Hedri +8
We present a general classification of simplified models that lead to dark matter (DM) coannihilation processes of the form DM + X SM + SM, where X is a coann…
Coupling--mass mapping of di-jet peak searches
Bogdan A. Dobrescu, Felix Yu
We study hypothetical gauge bosons that may produce dijet resonances at the LHC. Simple renormalizable models include leptophobic Z' bosons or colorons that have flavor-independent…
Large Language Models are Interpretable Learners
Ruochen Wang, Si Si, Felix Yu +3
The trade-off between expressiveness and interpretability remains a core challenge when building human-centric predictive models for classification and decision-making. While symbo…
Scalable In-context Ranking with Generative Models
Nilesh Gupta, Chong You, Srinadh Bhojanapalli +3
In-context Ranking (ICR) is an emerging paradigm for Information Retrieval (IR), which leverages contextual understanding of LLMs by directly incorporating the task description, ca…
HD-cos Networks: Efficient Neural Architectures for Secure Multi-Party Computation
Wittawat Jitkrittum, Michal Lukasik, Ananda Theertha Suresh +2
Multi-party computation (MPC) is a branch of cryptography where multiple non-colluding parties execute a well designed protocol to securely compute a function. With the non-colludi…
Quark-universal breaking scalar at the LHC
Lorin Armbruster, Bogdan A. Dobrescu, Felix Yu
If the quarks or leptons are charged under a new gauge symmetry, then besides a boson there must exist at least one new boson whose decay products include Standard Mode…
SPICE: Simulation Package for Including Flavor in Collider Events
Guy Engelhard, Jonathan L. Feng, Iftah Galon +2
We describe SPICE: Simulation Package for Including Flavor in Collider Events. SPICE takes as input two ingredients: a standard flavor-conserving supersymmetric spectrum and a set…
Autoregressive Ranking: Bridging the Gap Between Dual and Cross Encoders
Benjamin Rozonoyer, Chong You, Michael Boratko +5
The success of Large Language Models (LLMs) has motivated a shift toward generative approaches to retrieval and ranking, aiming to supersede classical Dual Encoders (DEs) and Cross…
A Tale of Two Portals: Testing Light, Hidden New Physics at Future Colliders
Jia Liu, Xiao-Ping Wang, Felix Yu
We investigate the prospects for producing new, light, hidden states at a future collider in a Higgsed dark model, which we call the Double Dark Portal model. Th…
Two-stage LLM Fine-tuning with Less Specialization and More Generalization
Yihan Wang, Si Si, Daliang Li +5
Pretrained large language models (LLMs) are general purpose problem solvers applicable to a diverse set of tasks with prompts. They can be further improved towards a specific task…
Primer on Axion Physics
Felix Yu
I review the canonical axion potential, with an emphasis on the field theory underlying radial and angular modes of complex scalar fields. I present the explicit calculation of the…
Semantic Label Smoothing for Sequence to Sequence Problems
Michal Lukasik, Himanshu Jain, Aditya Krishna Menon +4
Label smoothing has been shown to be an effective regularization strategy in classification, that prevents overfitting and helps in label de-noising. However, extending such method…
Effects from New Colored States and the Higgs Portal on Gluon Fusion and Higgs Decays
Kunal Kumar, Roberto Vega-Morales, Felix Yu
We study effects from new colored states and the Higgs portal on gluon fusion production. We isolate possible loop contributions from new colored scalars, fermions, and vectors, in…
Discovery potential of Kaluza-Klein gluons at hadron colliders: A Snowmass whitepaper
Kyoungchul Kong, Felix Yu
We investigate the discovery potential of Kaluza-Klein gluons as a dijet resonance at hadron colliders with different center-of-mass energies, from 14 TeV to 33 TeV to 100 TeV. We…
A Viable Flavor Model for Quarks and Leptons in RS with T' Family Symmetry
Mu-Chun Chen, K. T. Mahanthappa, Felix Yu
We propose a Randall-Sundrum model with a bulk family symmetry based on the double tetrahedral group, T', which generates the tri-bimaximal neutrino mixing pattern and a realistic…
Anatomizing Exotic Production of the Higgs Boson
Felix Yu
We discuss exotic production modes of the Higgs boson and how their phenomenology can be probed in current Higgs analyses. We highlight the importance of differential distributions…
LoRA Done RITE: Robust Invariant Transformation Equilibration for LoRA Optimization
Jui-Nan Yen, Si Si, Zhao Meng +5
Low-rank adaption (LoRA) is a widely used parameter-efficient finetuning method for LLM that reduces memory requirements. However, current LoRA optimizers lack transformation invar…
InFillmore: Frame-Guided Language Generation with Bidirectional Context
Jiefu Ou, Nathaniel Weir, Anton Belyy +2
We propose a structured extension to bidirectional-context conditional language generation, or "infilling," inspired by Frame Semantic theory (Fillmore, 1976). Guidance is provided…
Doubly-stochastic mining for heterogeneous retrieval
Ankit Singh Rawat, Aditya Krishna Menon, Andreas Veit +3
Modern retrieval problems are characterised by training sets with potentially billions of labels, and heterogeneous data distributions across subpopulations (e.g., users of a retri…
Hunting for Dark Matter Coannihilation by Mixing Dijet Resonances and Missing Transverse Energy
Malte Buschmann, Sonia El Hedri, Anna Kaminska +5
Simplified models of the dark matter (co)annihilation mechanism predict striking new collider signatures untested by current searches. These models, which were codified in the coan…
Measuring Slepton Masses and Mixings at the LHC
Jonathan L. Feng, Sky T. French, Iftah Galon +5
Flavor physics may help us understand theories beyond the standard model. In the context of supersymmetry, if we can measure the masses and mixings of sleptons and squarks, we may…
Supersymmetric Exotic Decays of the 125 GeV Higgs Boson
Jinrui Huang, Tao Liu, Lian-Tao Wang +1
We reveal a set of novel decay topologies for the 125 GeV Higgs boson in supersymmetry which are initiated by its decay into a pair of neutralinos, and discuss their collider searc…
Parameter Space of General Gauge Mediation
Arvind Rajaraman, Yuri Shirman, Joseph Smidt +1
We study a subspace of General Gauge Mediation (GGM) models which generalize models of gauge mediation. We find superpartner spectra that are markedly different from those of typic…
Take the Scenic Route: Improving Generalization in Vision-and-Language Navigation
Felix Yu, Zhiwei Deng, Karthik Narasimhan +1
In the Vision-and-Language Navigation (VLN) task, an agent with egocentric vision navigates to a destination given natural language instructions. The act of manually annotating the…
Correlated quantization for distributed mean estimation and optimization
Ananda Theertha Suresh, Ziteng Sun, Jae Hun Ro +1
We study the problem of distributed mean estimation and optimization under communication constraints. We propose a correlated quantization protocol whose leading term in the error…
A Viable Randall-Sundrum Model for Quarks and Leptons with T' Family Symmetry
Mu-Chun Chen, K. T. Mahanthappa, Felix Yu
We propose a Randall-Sundrum model with a bulk family symmetry based on the double tetrahedral group, T', which generates the tri-bimaximal neutrino mixing pattern and a realistic…
Simplified Models for LHC New Physics Searches
Daniele Alves, Nima Arkani-Hamed, Sanjay Arora +92
This document proposes a collection of simplified models relevant to the design of new-physics searches at the LHC and the characterization of their results. Both ATLAS and CMS hav…
Baby Bear: Seeking a Just Right Rating Scale for Scalar Annotations
Xu Han, Felix Yu, Joao Sedoc +1
Our goal is a mechanism for efficiently assigning scalar ratings to each of a large set of elements. For example, "what percent positive or negative is this product review?" When s…
Indirect Probes of the MSSM after the Higgs Discovery
Wolfgang Altmannshofer, Marcela Carena, Nausheen R. Shah +1
We study the minimal supersymmetric standard model (MSSM) with minimal flavor violation (MFV), imposing constraints from flavor physics observables and MSSM Higgs searches, in ligh…
Report of the Topical Group on Physics Beyond the Standard Model at Energy Frontier for Snowmass 2021
Tulika Bose, Antonio Boveia, Caterina Doglioni +318
This is the Snowmass2021 Energy Frontier (EF) Beyond the Standard Model (BSM) report. It combines the EF topical group reports of EF08 (Model-specific explorations), EF09 (More gen…
Angular observables for spin discrimination in boosted diboson final states
Malte Buschmann, Felix Yu
We investigate the prospects for spin determination of a heavy diboson resonance using angular observables. Focusing in particular on boosted fully hadronic final states, we detail…
Self-supervised Learning for Large-scale Item Recommendations
Tiansheng Yao, Xinyang Yi, Derek Zhiyuan Cheng +8
Large scale recommender models find most relevant items from huge catalogs, and they play a critical role in modern search and recommendation systems. To model the input space with…
Collider Signals of Maximal Flavor Violation: Same-Sign Leptons from Same-Sign Tops at the Tevatron
Shaouly Bar-Shalom, Arvind Rajaraman, Daniel Whiteson +1
In models of maximal flavor violation (MxFV) there is at least one new scalar which couples to the quarks via where $ξ_{i3},ξ_{3i} \si…
Axion Couplings in Gauged Extensions of the Standard Model
Alexey Kivel, Julien Laux, Felix Yu
We explore the effective theory of an axion in a gauged baryon number symmetry extension of the Standard Model (SM), where the axion is realized from a Dine-Fischler-Srednicki-Zhit…
Constraining RS Models by Future Flavor and Collider Measurements: A Snowmass Whitepaper
Kaustubh Agashe, Martin Bauer, Florian Goertz +4
Randall-Sundrum models are models of quark flavor, because they explain the hierarchies in the quark masses and mixings in terms of order one localization parameters of extra dimen…
Quantum Imprint of the Anharmonic Oscillator
Prisco Lo Chiatto, Sebastian Schenk, Felix Yu
We study the anharmonic double well in quantum mechanics using exact Wentzel-Kramers-Brillouin (WKB) methods in a 't Hooft-like double scaling limit where classical behavior is exp…
Efficient and Asymptotically Unbiased Constrained Decoding for Large Language Models
Haotian Ye, Himanshu Jain, Chong You +4
In real-world applications of large language models, outputs are often required to be confined: selecting items from predefined product or document sets, generating phrases that co…
Large Language Models with Controllable Working Memory
Daliang Li, Ankit Singh Rawat, Manzil Zaheer +5
Large language models (LLMs) have led to a series of breakthroughs in natural language processing (NLP), owing to their excellent understanding and generation abilities. Remarkably…
FedDM: Iterative Distribution Matching for Communication-Efficient Federated Learning
Yuanhao Xiong, Ruochen Wang, Minhao Cheng +2
Federated learning~(FL) has recently attracted increasing attention from academia and industry, with the ultimate goal of achieving collaborative training under privacy and communi…
Sampled Softmax with Random Fourier Features
Ankit Singh Rawat, Jiecao Chen, Felix Yu +2
The computational cost of training with softmax cross entropy loss grows linearly with the number of classes. For the settings where a large number of classes are involved, a commo…
Regression-aware Inference with LLMs
Michal Lukasik, Harikrishna Narasimhan, Aditya Krishna Menon +2
Large language models (LLMs) have shown strong results on a range of applications, including regression and scoring tasks. Typically, one obtains outputs from an LLM via autoregres…
Learning discrete distributions: user vs item-level privacy
Yuhan Liu, Ananda Theertha Suresh, Felix Yu +2
Much of the literature on differential privacy focuses on item-level privacy, where loosely speaking, the goal is to provide privacy per item or training example. However, recently…
New Physics Models of Direct CP Violation in Charm Decays
Wolfgang Altmannshofer, Reinard Primulando, Chiu-Tien Yu +1
In view of the recent LHCb measurement of Delta A_CP, the difference between the time-integrated CP asymmetries in D --> K+K- and D --> pi+pi- decays, we perform a comparative stud…
A Z' Model for the CDF Dijet Anomaly
Felix Yu
We adopt a bottom-up approach to constructing a new physics model to explain the CDF excess seen in dijets with an associated lepton and missing transverse energy. We find that the…
Measuring CP Violation in at Colliders
Roni Harnik, Adam Martin, Takemichi Okui +2
We investigate the LHC and Higgs Factory prospects for measuring the CP phase in the Higgs-tau-tau coupling. Currently this phase can be anywhere between 0 degrees (CP even) and 90…
Machine Learning on Heterogeneous, Edge, and Quantum Hardware for Particle Physics (ML-HEQUPP)
Julia Gonski, Jenni Ott, Shiva Abbaszadeh +118
The next generation of particle physics experiments will face a new era of challenges in data acquisition, due to unprecedented data rates and volumes along with extreme environmen…
Lattice Rescoring Strategies for Long Short Term Memory Language Models in Speech Recognition
Shankar Kumar, Michael Nirschl, Daniel Holtmann-Rice +3
Recurrent neural network (RNN) language models (LMs) and Long Short Term Memory (LSTM) LMs, a variant of RNN LMs, have been shown to outperform traditional N-gram LMs on speech rec…
Supersymmetric Sub-Electroweak Scale Dark Matter, the Galactic Center Gamma-ray Excess, and Exotic Decays of the 125 GeV Higgs Boson
Jinrui Huang, Tao Liu, Lian-Tao Wang +1
We continue our exploration of the nearly Peccei-Quinn symmetric limit shared by common singlet extensions of the Minimal Supersymmetric Standard Model. This limit has been establi…
Spark Transformer: Reactivating Sparsity in FFN and Attention
Chong You, Kan Wu, Zhipeng Jia +16
The discovery of the lazy neuron phenomenon in trained Transformers, where the vast majority of neurons in their feed-forward networks (FFN) are inactive for each token, has spurre…
Dijet and electroweak limits on a boson coupled to quarks
Bogdan A. Dobrescu, Felix Yu
An insightful way of presenting the LHC limits on dijet resonances is the coupling-mass plot for a boson that has flavor-independent quark interactions. This also illustrates…
A Lighter QCD Axion from Anarchy
Fatemeh Elahi, Gilly Elor, Alexey Kivel +3
We introduce the Anarchic Axion, a class of axion models which solves the Strong CP problem within current nEDM constraints with a lighter than usual QCD axion, thus populating new…
Three-Body Decays of Sleptons with General Flavor Violation and Left-Right Mixing
Jonathan L. Feng, Iftah Galon, David Sanford +2
We determine the widths of three-body decays of sleptons, , in the presence of arbitrary s…
Efficient Document Ranking with Learnable Late Interactions
Ziwei Ji, Himanshu Jain, Andreas Veit +6
Cross-Encoder (CE) and Dual-Encoder (DE) models are two fundamental approaches for query-document relevance in information retrieval. To predict relevance, CE models use joint quer…
Consistent Electroweak Phenomenology of a Nearly Degenerate Boson
Prisco Lo Chiatto, Felix Yu
Extracting constraints on kinetic mixing between a new gauge boson hiding under the Standard Model boson resonance requires the formalism of non-Hermitian two-point cor…
A New Method for Resolving Combinatorial Ambiguities at Hadron Colliders
Arvind Rajaraman, Felix Yu
We present a new method for resolving combinatorial ambiguities that arise in multi-particle decay chains at hadron colliders where the assignment of visible particles to the diffe…
Long-Lived Particles at the Energy Frontier: The MATHUSLA Physics Case
David Curtin, Marco Drewes, Matthew McCullough +85
We examine the theoretical motivations for long-lived particle (LLP) signals at the LHC in a comprehensive survey of Standard Model (SM) extensions. LLPs are a common prediction of…
Di-jet resonances at future hadron colliders: A Snowmass whitepaper
Felix Yu
I investigate the sensitivity of future hadron colliders to di-jet resonances arising from Z' or coloron models. The projected discovery potential and exclusion limits for these re…
Displaying dark matter constraints from colliders with varying simplified model parameters
Andreas Albert, Antonio Boveia, Oleg Brandt +17
The search for dark matter is one of the main science drivers of the particle and astroparticle physics communities. Determining the nature of dark matter will require a broad appr…
Hierarchical Retrieval: The Geometry and a Pretrain-Finetune Recipe
Chong You, Rajesh Jayaram, Ananda Theertha Suresh +3
Dual encoder (DE) models, where a pair of matching query and document are embedded into similar vector representations, are widely used in information retrieval due to their simpli…
Sensitivity of potential future colliders to quark compositeness
Leonard Apanasevich, Suneet Upadhyay, Nikos Varelas +2
A study is presented of the sensitivity of potential future colliders to quark compositeness. The analysis uses normalized dijet angular distributions compared to expectations…
Exotic Signals of Vectorlike Quarks
Bogdan A. Dobrescu, Felix Yu
Vectorlike fermions are an important target for hadron collider searches. We show that the vectorlike quarks may predominantly decay via higher-dimensional operators into a quark p…
SpecTr: Fast Speculative Decoding via Optimal Transport
Ziteng Sun, Ananda Theertha Suresh, Jae Hun Ro +3
Autoregressive sampling from large language models has led to state-of-the-art results in several natural language tasks. However, autoregressive sampling generates tokens one at a…
Supersizing axions with small size instantons
Alexey Kivel, Julien Laux, Felix Yu
We construct a new framework to calculate the enhancement of axion masses and concomitant effects on axion-meson mixing arising from small size instantons (SSIs), which originate i…