Excited to share that Bisecle is accepted at #NeurIPS2025.
🧠 Bisecle: Binding and Separation in Continual Learning for Video Language Understanding.
Preprint: arxiv.org/abs/2507.00469
Code: github.com/cruiseresear...
Inspired by the rapid binding and pattern separation mechanisms in the hippocampus

Beyond Words: Integrating Theory of Mind into Conversational Agents for Human-Like Belief, Desire, and Intention Alignment (#ACL2025 Findings)

aclanthology.org/2025.finding...

Codes: github.com/cruiseresear...

Findings in the thread below.

NVIDIA Dynamo. Exciting announcement by Jensen Huang #gtc2025. A new open source VMWare-like inference framework for reasoning. It breaks up prefill and decode steps efficiently. DeepSeek R1 requests can be boosted by 30x. Already used by Perplexity, Meta, etc
developer.nvidia.com/blog/introdu...

The Building TimeSeries (BTS) dataset offers a rich, hierarchical structure of a large collection of diverse timeseries. The dataset include the following components:
* Multivariate TimeSeries: From 2021 to 2024, contains over 15,000 timeseries (TS) across 300 unique classes.
* Knowledge Graph
🧵

Another one down
BTS: Buildings Time Series
+ CRUISE in NeurIPS team photo
#neurips2024
@neuripsconf.bsky.social

One down
#neurips2024

My pick was Transformer! What about you?
@neuripsconf.bsky.social

Building TimeSeries (BTS) dataset covers 3 buildings over 3yrs, >10K timeseries data points with hundreds of unique ontologies. The metadata is standardized using the Brick schema (> 240 classes). We performed benchmarks on 2 tasks: timeseries classification and zero-shot forecasting.
#NeurIPS2024

ViLCo dataset: 10-min videos & language queries. ViLCo deals with memory complexity from long video clips, natural language complexity from open queries, and text-video misalignment, through a novel memory-efficient framework that incorporates SSL and long- & short-term memory.
#NeurIPS2024

Check out our 3 NeurIPS 📌 in thread.
1. ViLCo-Bench: VIdeo Language COntinual learning Benchmark
🎥 neurips.cc/virtual/2024...

2. BTS: Building Timeseries Dataset
🎥 neurips.cc/virtual/2024...

3. Resolution-Agnostic Transformer-based Climate Downscaling
🎥 neurips.cc/virtual/2024...

#NeurIPS2024

Fun times yesterday celebrating my student Yonchanok Khaokaew’s PhD graduation with some of my team members

I gave a keynote on LLMs (or not!): foundational approaches for timeseries and spatiotemporal data in SIGSPATIAL 2024, exec briefings this week in Cisco Live, and will give another keynote at AJCAI 2024 in 2wks.
I’ll be in NeurIPS next month to present 2 papers. Hope to cross path with some of you.