Sentence decoding

Name: sentence

Category: cognitive decoding

Dataset: Hollenstein2018 (ZuCo)

Objective: Retrieval

Split: Leave-subjects-out

Usage

neuralbench eeg sentence

Show config.yaml

# Copyright (c) Meta Platforms, Inc. and affiliates.
# All rights reserved.
#
# This source code is licensed under the license found in the
# LICENSE file in the root directory of this source tree.

data:
  study:
    source:
      name: Hollenstein2018Zuco
    split:
      name: SklearnSplit
      split_by: text
      valid_split_ratio: 0.1
      test_split_ratio: 0.2
      valid_random_state: 33
      test_random_state: 33
  channel_positions:
    layout_or_montage_name: GSN-HydroCel-128
  target:
    name: HuggingFaceText
    event_types: Sentence
    contextualized: false
    model_name: "openai-community/gpt2"
    layers: 0.6667
    token_aggregation: mean
    aggregation: trigger
  trigger_event_type: Sentence
  start: 0.0
  duration: 3.0
  stride: 3.0
  stride_drop_incomplete: false
  summary_columns: [task, text, sentence]
brain_model_output_size: &brain_model_output_size 768
target_scaler:
  dim: 1
trainer_config.monitor: val/batch_top5_acc
trainer_config.mode: max
loss:
  name: ClipLoss
  norm_kind: y
  temperature: false
  symmetric: false
metrics: !!python/name:neuralbench.defaults.metrics.retrieval_metrics
test_full_retrieval_metrics: !!python/name:neuralbench.defaults.metrics.test_full_retrieval_metrics

Description

The sentence decoding task involves decoding sentence stimuli from EEG segments. In this task, we use the ZuCo dataset [Hollenstein2018], which contains EEG data recorded while subjects read sentences presented one at a time on a screen.

Dataset Notes

• To avoid sentence leakage, the SR task recordings are used for testing, while the NR and TSR task recordings are concatenated and used for training and validation.

• To ensure a large enough number of examples for training, sliding window segments are extracted within each sentence presentation, and are mapped to the same sentence representation.

References

[Hollenstein2018]

Hollenstein, Nora, et al. “ZuCo, a simultaneous EEG and eye-tracking resource for natural sentence reading.” Scientific data 5.1 (2018): 1-13.