@ShaLabUSC

Machine learning lab at @USC, led by @feishaAI.

 Los Angeles, CA
Joined February 2018
  • Tweets22
  • Following11
  • Followers94
5Media
Excited to share our work in collaboration with @whi_rl! We propose a multi-agent RL technique to improve generalization across scenarios where the types and quantities of agents change by "imagining" sub-scenarios. arxiv.org/abs/2006.04222
2
8
2
40
ShaLab retweeted
Kyle Wiggers
@Kyle_L_Wiggers12 May 2020
BabyWalk AI breaks complex navigation into simple steps ift.tt/2LmtJRV
7
5
It is out ! @ShaLabUSC @feishaAI
Stat.ML Papers@StatMLPapers5 Feb 2019
Hyper-parameter Tuning under a Budget Constraint. (arXiv:1902.00532v1 [cs.LG]) bit.ly/2SqVbD3

Hyper-parameter Tuning under a Budget Constraint

We study a budgeted hyper-parameter tuning problem, where we optimize the tuning result under a hard resource constraint. We propose to solve it as a sequential decision making problem, such that...

arxiv.org
1
1
1
Our lab’s work on zero-shot learning, combining several conference papers and updating with new perspectives, results and evaluation metrics, along with codes. ⁦@ShaLabUSC⁩ ⁦@weilunchao⁩ ⁦@BoqingGo⁩ ⁦@martbeerina⁩ arxiv.org/abs/1812.06423

Classifier and Exemplar Synthesis for Zero-Shot Learning

Zero-shot learning (ZSL) enables solving a task without the need to see its examples. In this paper, we propose two ZSL frameworks that learn to synthesize parameters for novel unseen classes....

arxiv.org
3
1
14
Great article published this week by @feishaai about #Learning #ComputerScience #ComputerVision #PatternRecognition deepai.org/publication/learn…

Learning Embedding Adaptation for Few-Shot Learning

12/10/18 - Learning with limited data is a key challenge for visual recognition. Few-shot learning methods address this challenge by learning...

deepai.org
3
1
9
Prof. Marius Kloft, my friend who has been spending his sabbatical in my lab, is guest lecturing statistical learning theory. :)
1
16
Hey! Our NEW zero-shot learning toolbox has been released: github.com/pujols/Zero-shot-…. Includes our algorithms SynC and EXEM, as well as scripts for generalized zero-shot learning evaluation. Joint work with @weilunchao @boqinggo and @feishaAI.

GitHub - pujols/Zero-shot-learning-journal

Contribute to pujols/Zero-shot-learning-journal development by creating an account on GitHub.

github.com
5
12
How a bot (or perhaps even a person) can be an engaging conversation partner? - being curious! arxiv.org/abs/1808.07104 to appear at @conll2018 @ShaLabUSC
4
7
Is multi-task learning useful in NLP? Our (@shalabusc) extensive empirical study appearing at @coling2018 suggests so, at least half of the time for sequence tagging tasks. arxiv.org/abs/1808.04151
5
1
12
Listened to soon-to-be Assistant Professor Wei-Lun (Harry) Chao's dry run for a 100th time. Great work!!!
3
2
A subset of @ShaLabUSC enjoyed hot pots after #SoCalNLP. Thanks to the organizers for the great event!
2
4
Load more