Search is not available for this dataset
spike_times unknown | shape listlengths 3 3 | num_steps int32 100 100 | label class label 1.65k
classes |
|---|---|---|---|
[
54,
56,
57,
56,
55,
53,
52,
53,
52,
53,
55,
56,
56,
56,
57,
56,
58,
61,
61,
61,
61,
60,
62,
61,
60,
61,
63,
62,
63,
64,
67,
66,
65,
65,
66,
67,
66,
66,
68,
68,
69,
70,
71,
73,
73,
73,
74,
75,
75,
76,
75,
74,
74,
... | [
3,
128,
128
] | 100 | 000001_aardvark |
[
68,
68,
66,
68,
66,
64,
65,
68,
68,
68,
67,
66,
65,
63,
64,
65,
66,
66,
67,
66,
66,
67,
67,
65,
66,
68,
68,
66,
65,
65,
65,
65,
64,
63,
61,
61,
63,
62,
61,
62,
61,
61,
61,
61,
61,
62,
63,
63,
63,
65,
66,
63,
61,
... | [
3,
128,
128
] | 100 | 000001_aardvark |
"OjpASlVYVks0LCgoKi0wLy8vLi4vMDEyNDc2NjU1LCY5VllYWVlZWVlZWVlZWVpaWlpZWVlZWlpZVlFEMy0sKiwoJCYmJiMlJiY(...TRUNCATED) | [
3,
128,
128
] | 100 | 000001_aardvark |
"QUNDRURFRUdHOzQ2PEJJSUhGRERCPUFEQkJKTElGQkBDRkdFREZGQ0RFRkZAREpHRERFRUFCRUREREJCPz5AQUFDRENESUdEREZ(...TRUNCATED) | [
3,
128,
128
] | 100 | 000001_aardvark |
"R0lKTE1LTk5OTUxOS01QUFJVVVJRUk0zQkhNUlNSVVZSUVNSUFFSU1JTUlJUUlBQUFFPT1JTUE5MT1BQTkxOTVFTUlBMTE5WWVl(...TRUNCATED) | [
3,
128,
128
] | 100 | 000001_aardvark |
"YGFhYGBgYGFhYWFbUkZCR1JdYGBgYGFgYF9eXl9gXl5eXl5eXl5eXl5eXl5eXV1dXV1dXFxdXVtbW1tbWlpaW1tcWlpbW1xbWll(...TRUNCATED) | [
3,
128,
128
] | 100 | 000001_aardvark |
"XVxYWFpaV1pdXV5gXl1gX15eXVxbWVhbXl5cXVpZV1dTUlVZXFxcXFpZWlxfYF1aW1xaXFtZWlxbXV5dXF1eXl9eXFxeX1xaX1x(...TRUNCATED) | [
3,
128,
128
] | 100 | 000001_aardvark |
"PkBDSEc3MSwoKisoKz09PTw4OTY0NjU0NDU3Ojk3NzY4Nzg2NzQqIyQmKCsuLS8tKi0wMC8vLSsqLS4uLy8xMTAvMzlCQEBHS0x(...TRUNCATED) | [
3,
128,
128
] | 100 | 000001_aardvark |
"HyMnKigkIx8fJCMiJSgqJh0TDRATFhgZFREQERYgIRwbHSEcICgiGRogJCgoKScoLCsqKyYiJCQoKCcqJR8oKCEfIx8YFBASGB0(...TRUNCATED) | [
3,
128,
128
] | 100 | 000001_aardvark |
"QklVWFFCSFBSUlJSUUo5P0xORzopJCw0ODo3MTg/NCYiIR8fHyIlKzQ9QUJGSklFPzo6P0dIRUFBQkNCPTYpIB4QCwkgGAMEAwY(...TRUNCATED) | [
3,
128,
128
] | 100 | 000001_aardvark |
End of preview. Expand in Data Studio
This dataset is spike encoded data of the THINGS EEG. This data is latency coded (latency coding uses input features to determine spiking timing). The num of steps is 100. The image resolution was reduced from 224 to 128 and features standardized. threshold=0.01, Tau=5.
To reconstruct the images run the script below.
import matplotlib.pyplot as plt
import numpy as np
from datasets import load_dataset
REPO_ID = "corquaerit/things-eeg-spikes-LC"
NUM_STEPS = 100
TAU = NUM_STEPS - 1
spike_ds = load_dataset(REPO_ID, split="test", streaming=True)
label_feature = spike_ds.features["label"]
def unpack_row(row):
spike_times = np.frombuffer(row["spike_times"], dtype=np.uint8).reshape(row["shape"])
never_fired = spike_times == 255
pixel_values = 1.0 - (spike_times.astype(np.float32) / TAU)
pixel_values[never_fired] = 0.0
return np.clip(pixel_values, 0.0, 1.0) # [C, H, W]
iterator = iter(spike_ds)
n_examples = 3
fig, axes = plt.subplots(1, n_examples, figsize=(4 * n_examples, 4))
for i in range(n_examples):
row = next(iterator)
reconstructed = unpack_row(row)
reconstructed = np.transpose(reconstructed, (1, 2, 0))
label_name = label_feature.int2str(row["label"])
axes[i].imshow(reconstructed)
axes[i].set_title(label_name, fontsize=9)
axes[i].axis("off")
plt.tight_layout()
plt.show()
- Downloads last month
- 31