Errata: September 12, 2022Thank you for purchasing Deep Learning with PyTorch . Please post errata not listed below in this book's LiveBook Errata thread. All formats are corrected during the second printing. Thank you!
The following corrections were made during the book's reprint in June 2022.
Chapter 10, Section 10.2 Parsing LUNA's annotation data
, page 288
After Figure 10.3, in Paragraph 2: ...for a candidate that is not an actual nodule, and 1 for a candidate that is a nodule, either malignant or benign). We have 1,351 candidates flagged as actual nodules.
should be ...for a candidate that is not a nodule, e.g. a blood vessel, and 1 for a candidate that is a nodule, which can be malignant or benign). We have 1,351 candidates flagged as nodules.
Chapter 10, Section 10.5 A straightforward dataset implementation
, page 272
In Listing 10.12: Delete the two text snippets 1((CO10-1))
and 1((CO10-2))
from the Listing.
Chapter 10, Section 10.5.4 Rendering the data
, page 277
In the first code annotation, delete the annotation and it should read:
# In[7]:
from p2ch10.vis import
findPositiveSamples, showCandidate
positiveSample_list = findPositiveSamples()
The following corrections were made during the book's reprint in Winter 2021.
Page xxviii
The "about this cover" section incorrectly identifies the individual on the cover as Kardinian, but he is Kabardian.
Page 5
Figure 1.1 was incorrectly flipped left-to-right.
Page 195
The third code snippet incorrectly referenced "i22" instead of "i13" in the first line. The full listing should be:
o11 = i11 * w00 + i12 * w01 + i13 * w02 +
i21 * w10 + i22 * w11 + i23 * w12 +
i31 * w20 + i32 * w21 + i33 * w22
Page 205
The reference to a "MaxPool3d" should be "MaxPool2d".
Page 252
The final paragraph of section 9.4.4 should read:
Once you have the candidates file and at least one subset downloaded, uncompressed, and put in the correct location, you should be able to start running the examples in the next chapter. If you want to jump ahead, you can use the p2ch10_explore_data.ipynb Jupyter Notebook to get started. Otherwise, we'll return to the notebook in more depth in the next chapter. Hopefully your downloads will finish before you start needing to use the data!
Page 258
The first code annotation should read:
Counts the number of lines that end with 1, which indicates it is a nodule.
The end of the first paragraph should read:
(it's a Boolean value: 0 for a candidate that is not a nodule, e.g. a blood vessel, and 1 for a candidate that is a nodule, which can be malignant or benign). We have 1,351 of the nodule candidates flagged as nodules.