xianyicheng/stage_classification

This is the visualization of our stage classification result for our screwdriving project.

---

Anomaly detection:

In the case where all the stages in an operation are classified as 'approach', the system would actually specified this operation as 'anomaly'. Also, the operation with no stage label but with 'unknown' as its tiltle, is also an 'anomaly'. These two case are actually the same. This is caused by some inconsistency of plot labeling code across our different test. We will fix this.

---

Below is the description about each folder：

---

dataset1

The model HMM1_before, with parameter estimation of labeled samples, produce the following stage classification:

before1_accurate: dataset1; accurate alignment screwdriving operations, before HMM adaptation

before1_error: dataset1; injected alignment error screwdriving operations, before HMM adaptation

The model HMM1_after, perform data adaptation on 50 unlabeled samples, produce the following stage classification:

after1_accurate: dataset1; accurate alignment screwdriving operations, after HMM adaptation

after1_error: dataset1; injected alignment error screwdriving operations, after HMM adaptation

---

dataset 2

The model HMM2_before, with just modification of experimental parameters of HMM1_before, produce the following stage classification:

before2: dataset2

The model HMM2_after, perform data adaptation on 50 unlabeled samples based on HMM2_before, produce the following stage classification:

after2: dataset2

---

dataset 3

The model HMM3_before, with just modification of experimental parameters of HMM1_before, produce the following stage classification:

before3

The model HMM3_after, perform data adaptation on 75 unlabeled samples based on HMM3_before, produce the following stage classification:

after3

---

dataset 4

The model HMM4_before, with just modification of experimental parameters of HMM1_before, produce the following stage classification:

before4

The model HMM4_after, perform data adaptation on 10 unlabeled samples based on HMM4_before, produce the following stage classification:

after4

stage_classification's Issues

Study anomaly

dear @XianyiCheng,
How are you?
It is happy to read your master for me and can i test your code?