Hi! 你好We are the DeCLaRe Lab at Nanyang Technological University

Explore our research

Selected Projects

All projects

2017–2023

Landmark work

Tensor Fusion Network comparing early concatenation with an outer product containing unimodal, bimodal and trimodal regions

Multimodality · 2017

Tensor Fusion Network (TFN)

TFN represents every unimodal, bimodal and trimodal interaction among language, facial behaviour and voice with one outer product. The paper tests whether that explicit structure improves sentiment prediction over concatenation.

Shared implementation collection · 921 stars
Methods and results
DialogueRNN architecture tracking a global conversation state, separate speaker states and emotion representations across turns

Multimodality · 2018–2019

MELD + DialogueRNN

MELD pairs 13,708 conversational utterances with text, audio, video, emotion and sentiment labels. DialogueRNN tests whether tracking each speaker separately improves emotion classification.

MELD · 1,079 stars 1,944/mo
Methods and results
The TangoFlux training pipeline, showing text and audio encoding, flow-matching pre-training and preference alignment

Multimodality · Efficiency · 2023–2026

TANGO

Across five systems, this work moves from conditioning short sound generation on language, to learning event order from preference pairs, exposing musical controls, reducing sampling time and placing individual words inside full songs.

Tango · 1,238 stars 50/mo
Methods and results

2025–2026

Current directions

A ScrambleToolBench episode in which an agent tests obfuscated tools, observes errors and clues, and carries what it learned into the next task

Multimodality · Efficiency · 2026

ScrambleToolBench + MNIST-PRO

ScrambleToolBench measures whether agents can identify and reuse obfuscated tools as their mappings change. MNIST-PRO measures whether vision-language models can assemble movable image glimpses into a correct spatial state.

ScrambleToolBench · 4 stars
Methods and results
The delta-mem architecture, with an 8 by 8 online state producing low-rank corrections to attention in a frozen language model

Efficiency · Trustworthiness · 2026

δ-mem + Σ-Mem

δ-mem writes prior context into a fixed 8 × 8 state that steers a frozen language model. Σ-Mem records task-conditioned peer correctness and co-error patterns for multi-agent selection, routing and voting.

δ-mem · 255 stars
Methods and results
KAIROS benchmark varying rapport history, peer behaviour and model confidence before a multi-agent answer

Trustworthiness · Safety · 2025–2026

KAIROS + ECL

KAIROS varies prior rapport, current peer behaviour and model confidence across 3,000 questions. Epistemic Context Learning makes an agent infer who has been reliable before it sees their advice on the next question.

KAIROS · 8 stars 199/mo
Methods and results

Lab Notes

All notes

Subscribe by RSS

Code, Models and Data

More releases

Recent highlights

All updates
Paper

ScrambleToolBench: adapting to hidden tool changes

A terminal benchmark for agents that must infer undocumented tool behaviour and recover when the tool-to-function mapping changes.

Read the paper
Achievement

Research internships for DeCLaRe students

Four students started internships at Tencent, MiniMax and Meta.

Meet the team
Media

δ-mem featured in VentureBeat

The article covers δ-mem's compact online state and its low-rank interface to a frozen language model.

Read the coverage
Grant

Embodied Foundational Models funded by CNRS@CREATE and NRF

A 2026–2029 project on generalist vision-language-action models for embodied AI.

Funding details

People at DeCLaRe

Faculty, research staff and students at NTU Singapore.

  • Soujanya Poria
  • Suorong Yang
  • Rishabh Bhardwaj
  • Navonil Majumder
  • Soumitra Sinhahajari
  • Jingdi Lei

About DeCLaRe

DeCLaRe, short for Deep Cognition and Language Research, was founded by Soujanya Poria at the Singapore University of Technology and Design in 2019 with Navonil Majumder, Devamanyu Hazarika and Deepanway Ghosal. The lab moved to Nanyang Technological University in 2025.

Lab's identity

The robot forms 宣 (xuān), “to declare,” carrying the lab's name directly into its visual identity.

宣 · Xuān The robot is shaped from 宣, meaning “to declare.”