-
ARISE: Iterative Rule Induction and Synthetic Data Generation for Text Classification
Authors:
Yashwanth M.,
Vaibhav Singh,
Ayush Maheshwari,
Amrith Krishna,
Ganesh Ramakrishnan
Abstract:
We propose ARISE, a framework that iteratively induces rules and generates synthetic data for text classification. We combine synthetic data generation and automatic rule induction, via bootstrapping, to iteratively filter the generated rules and data. We induce rules via inductive generalisation of syntactic n-grams, enabling us to capture a complementary source of supervision. These rules alone…
▽ More
We propose ARISE, a framework that iteratively induces rules and generates synthetic data for text classification. We combine synthetic data generation and automatic rule induction, via bootstrapping, to iteratively filter the generated rules and data. We induce rules via inductive generalisation of syntactic n-grams, enabling us to capture a complementary source of supervision. These rules alone lead to performance gains in both, in-context learning (ICL) and fine-tuning (FT) settings. Similarly, use of augmented data from ARISE alone improves the performance for a model, outperforming configurations that rely on complex methods like contrastive learning. Further, our extensive experiments on various datasets covering three full-shot, eight few-shot and seven multilingual variant settings demonstrate that the rules and data we generate lead to performance improvements across these diverse domains and languages.
△ Less
Submitted 9 February, 2025;
originally announced February 2025.
-
Measurements of thermal relaxation of the OGRAN underground setup
Authors:
Gavrilyuk Y. M.,
Gusev A. V.,
Kvashnin N. L.,
Lugovoy A. A.,
Oreshkin S. I.,
Popov S. M.,
Rudenko V. N.,
Semenov V. V.,
Syrovatsky I. A
Abstract:
An upgraded version of the OGRAN -- combined optical-acoustic gravitational wave detector -- has been investigated in a long-term operation mode. This installation, located at the Baksan Neutrino Observatory (BNO) INR RAS, is designed to work under the program for detecting collapsing stars in parallel with the neutrino detector: Baksan Underground Scintillation Telescope (BUST). Such joint search…
▽ More
An upgraded version of the OGRAN -- combined optical-acoustic gravitational wave detector -- has been investigated in a long-term operation mode. This installation, located at the Baksan Neutrino Observatory (BNO) INR RAS, is designed to work under the program for detecting collapsing stars in parallel with the neutrino detector: Baksan Underground Scintillation Telescope (BUST). Such joint search corresponds to the modern trend for a development of "multi-messenger astronomy". In this work the effects of thermal relaxation OGRAN are experimentally investigated using passive and active thermal stabilization systems in the underground laboratory BNO PK-14.
△ Less
Submitted 31 October, 2021;
originally announced November 2021.
-
Conversion and storage of modes with orbital angular momentum in quantum memory scheme
Authors:
Vashukevich E. A.,
Golubeva T. Yu.,
Golubev Yu. M
Abstract:
The paper studies the Raman quantum memory protocol as applied to quantum light with orbital angular momentum. The memory protocol is implemented on an ensemble of three-level cold atoms with the $Λ$- configuration of energy levels. The possibility of storing quantum statistics of light with an orbital momentum is analysed in the case when the driving field could be treated as a plane wave. The ef…
▽ More
The paper studies the Raman quantum memory protocol as applied to quantum light with orbital angular momentum. The memory protocol is implemented on an ensemble of three-level cold atoms with the $Λ$- configuration of energy levels. The possibility of storing quantum statistics of light with an orbital momentum is analysed in the case when the driving field could be treated as a plane wave. The efficiency analysis shows that examined storage/retrieval processes do not cause the efficiency decreasing compared with the spatial multimode memory protocol considered in [Golubeva et al. 2012]]. We also present an effective transformation of the orbital angular momentum of a quantum field on a memory cell using the driving field with orbital angular momentum.
△ Less
Submitted 7 December, 2019;
originally announced December 2019.
-
Finding the optimal cluster state configuration. Cluster states classification by type of computations
Authors:
Korolev S. B.,
Golubeva T. Yu.,
Golubev Yu. M
Abstract:
In this paper, we study the transformations that are obtained in one-way quantum computation on continuous-variable cluster states of various configurations. Of all possible cluster configurations, we choose those that are suitable for universal Gaussian operations.
In this paper, we study the transformations that are obtained in one-way quantum computation on continuous-variable cluster states of various configurations. Of all possible cluster configurations, we choose those that are suitable for universal Gaussian operations.
△ Less
Submitted 26 November, 2019;
originally announced November 2019.
