WTF am I doing? An introduction to NLP and ANN's
summary
Project records active window titles once per minute and saves screen snapshots as 100x100 thumbnails, producing about 300,000 observations. Text uses bag‑of‑words and tf‑idf; images start with color histograms and then leverage pretrained CNN layers (VGG/ResNet) for feature vectors. Labels are hand‑annotated then filled by nearest-neighbor auto-labeling; prediction experiments use a simple MLP but remain inconclusive because labeling is noisy. Practical notes: storage and load times are manageable on an 8‑core i7.
AI-written summary · about
chapters (10)
- 0:00Personal NLP project intro
- 3:13Behavioral data collection
- 8:13Supervised vs unsupervised learning
- 11:05Labeling and feature challenges
- 14:41Dimensionality reduction
- 17:30Visualizing data clusters
- 20:21Image thumbnail labeling
- 22:39Pre-trained neural networks
- 24:28Sequence prediction
- 26:38Q&A
related talks
22mDeep learning module recognizes video sounds
43mGetting started with AI for free
01h 30mIntroduction to Deep Learning & Natural Language Processing
45mML: ML.NET: deep dive into dev AI
44mTeaching Computers ABCs: A Quick Intro to Natural Language Processing
30mStarting deep learning projects without enough labelled data
40mUse AI for "real things" in your Windows Apps
21mFrom 0 to a crowd sourced neural network in a jiffy