Spoken language identification system for detecting coastal karnataka languages using transfer learning and multi-view learning
Bibliographic record
Abstract
Abstract This paper presents approaches to develop spoken Language Identification (LID) system for identifying the three low-resource Indian languages - Kannada, Konkani, and Tulu, which are commonly spoken in the coastal region of Karnataka state of India. To address the challenges arising due to low-resource conditions, the proposed work aims to use a combination of data augmentation, transfer-learning and Multi-view learning. Specifically, noise perturbation and speed perturbation are used for data augmentation, and pre-trained Wav2Vec 2.0 and Whisper models are used for feature extraction, using which different Deep Learning (DL) based end-to-end models are trained for LID. Following this, a Multi-view learning based strategy is incorporated under which the LID model processes the feature representations obtained from Wav2Vec 2.0 and Whisper models simultaneously, using two separate input arms to capture the complimentary contents in them, leading to improved performance. Additionally, a combination of traditional Machine Learning (ML) with DL models is explored, in which, utterance-level embeddings obtained using pre-trained LID models are classified using separate back-end classifiers such as K-Nearest Neighbor (KNN) and Support Vector Machine (SVM). The results obtained highlight the advantage of using transfer-learning, Multi-view learning, and combination of DL-based model with ML-based classifiers to improve the overall performance of the LID system, amid low-resource settings. Specifically, combination of SVM backend on x-vector model with multi-view provided the best result compared to other models.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".