Pass the Fortinet NSE 6 Network Security Specialist NSE6_FSM_AN-7.4 Questions and answers with Dumpstech

The attribute must be selected as a Triggered Attribute so that it becomes available for incident generation and incident-title substitution. In the FortiSIEM Study Guide’s rule action configuration section, FortiSIEM separates incident attributes from triggered attributes. Triggered attributes are taken from the events that cause the rule to trigger and are then available for incident display and incident context. The guide explains that the rule action step is where an analyst defines the incident generated by a rule and chooses which attributes are carried forward. If Destination Host Name is not selected in the Triggered Attributes list, FortiSIEM cannot use it as an incident attribute in the generated incident title. Option B is wrong because aggregate items are used for calculations such as COUNT, AVG, or SUM, not for making a text attribute available in the incident title. Option C is wrong because Destination Host Name is an event attribute, not an event type. Option D is unrelated; removing Destination IP would not make Destination Host Name selectable.

The user initiates an RDP session (Subpattern 1) and then fails to log in multiple times (Subpattern 2 with COUNT(Matched Events) > = 3) - both from the same Source IP and User within 300 seconds.

The brute force attempts typically involve a successful RDP connection followed by multiple failed logins, satisfying the sequence and grouping conditions in the rule.

The correct answers are A and B because the rule uses multiple subpatterns and requires them to occur in a defined relationship within the configured time window. The FortiSIEM Study Guide states that rule conditions specify the event attributes and thresholds that trigger the rule and create an incident. It also explains that the time window is the period “within which the subpattern(s) must match for the rule condition to be satisfied,” and that when there is more than one subpattern, FortiSIEM requires logic between the subpatterns plus subpattern relationship constraints. In the exhibit, the first subpattern detects an RDP connection and the second detects failed logons. The rule condition uses a sequence relationship, so the failed logon activity must follow the RDP connection and match the relationship constraints, such as same user and source IP. A user using RDP over SSL VPN who fails repeatedly satisfies this logic. A brute-force attack against an RDP server also satisfies the repeated failed-logon requirement after an RDP connection. Failing only twice does not meet the aggregate threshold, and connecting to the wrong IP is not the failed-logon sequence being detected.

The exhibit shows a machine learning Regression configuration. In FortiSIEM regression, selected numerical fields are used as predictors, and another field is selected as the value to predict. The FortiSIEM 7.4 User Guide’s Machine Learning section lists Regression separately and describes the training workflow, including selecting fields used by the algorithm. The exhibit shows Memory Utilization, Sent Bytes, and Received Bytes selected under Fields to use for Prediction, while CPU Utilization is selected under Field to Predict. This means Memory Utilization is an input variable used by the model. If memory utilization is consistently high during training or retraining, that higher value becomes part of the learned model pattern. FortiSIEM does not automatically trigger a high-memory incident merely because a regression input value is high; an incident would require inference behavior and an anomaly action or rule configuration. Option C is also wrong because FortiSIEM does not dynamically lower a CPU trigger threshold just because memory is high. The correct behavior is model update based on the higher observed input value.

The automation policy is configured to run a remediation script named " Fortinet FortiOS - Block Source IP FortiOS via API " . It specifies enforcement on two FortiGate devices: FortiGate508 and FortiGate90D. Therefore, associated source IP addresses will be blocked on those two FortiGate firewalls only.

The correct answer is D because the remediation configuration defines specific enforcement targets. The FortiSIEM Study Guide explains that automation policies can run remediation scripts automatically when an incident occurs. It also explains the remediation options: Enforce On determines which devices the script runs against, while Run On identifies whether the script is launched from the supervisor or a collector. The Study Guide further states that mitigation scripts can block an IP address in a firewall or disable a user in Active Directory, and recommends specifying the Enforce On value because it controls the target device used by the remediation script. In the exhibit, the selected script is a Fortinet FortiOS block-source-IP remediation script, and the Enforce On field lists two FortiGate devices. That means the block action is targeted only at those two named FortiGate firewalls. The Aviation organization limits the automation policy context, but it does not mean every device in the organization receives the block. It is also not all FortiGate firewalls or the whole Network CMDB group.

The correct answer is C. Regression. Regression is the machine learning task used when a model predicts a target value based on other fields in a dataset. In FortiSIEM’s machine learning workflow, regression configuration includes selecting Fields to use for Prediction and a Field to Predict. The User Guide explains that during regression training, the analyst chooses the fields used for prediction and the field being predicted, then trains the model with a Train factor. This matches the question exactly: the model predicts a target based on other fields, and the inference phase can identify anomalies when observed values differ significantly from predicted values. Classification is used to assign records to discrete classes or categories, not to predict a continuous target metric from other variables. Clustering groups similar observations without a predefined target field. Forecasting predicts future values over time, normally based on time-series behavior. Because the question specifically describes predicting a target field from other dataset fields and detecting mismatch from the prediction, the correct machine learning method is Regression.

The operator is set to " = " , which performs an exact match on the entire raw event log, not a substring search. To find logs that contain the keyword " udp " , the analyst should use the CONTAIN operator instead. This will return all logs where " udp " appears anywhere in the raw log message.

The correct answer is D because the analyst is trying to search for raw logs that contain the keyword udp, but the filter uses the equality operator. The FortiSIEM Study Guide explains keyword searches in terms of Raw Event Log CONTAIN logic. In the keyword phrase search section, the guide states that without quotes, FortiSIEM searches raw event logs by using conditions such as Raw Event Log CONTAIN TCP OR Raw Event Log CONTAIN connection . Another analytics example explains that searching for TCP or UDP events uses the raw event log containing the keyword tcp or udp and that the search returns case-insensitive results for TCP and UDP. This directly eliminates option C. The time range is already set to the last two hours, which is correct for historical raw log searching. The problem is the operator. To find a keyword anywhere inside a raw log message, the analyst should use CONTAIN , not =.

The correct answer is C because the rule’s Group By attributes determine how events are grouped before the aggregate condition is evaluated. The Study Guide explains that rule conditions are built from subpatterns consisting of event attribute filters and aggregation functions. It also explains that a subpattern combines filters, aggregate, and group by fields to form the rule logic. In this case, the filters may return matching events in Analytics, but the rule still may not trigger because the aggregate condition is calculated separately for each unique Group By combination. The exhibit groups by Destination IP and User while applying COUNT(Source IP) > = 2. This means FortiSIEM does not count all matching events together. Instead, it counts only events that share the same Destination IP and User combination. If no single grouped combination reaches the aggregate threshold, no incident is created. The issue is not the event lookup, not the Destination Host Name format, and not necessarily the aggregate expression itself. The grouping logic is what restricts the counted event set.

The correct answer is A. FortiSIEM agent. The FortiSIEM Study Guide identifies FortiSIEM agents as the component responsible for “file, log monitoring, and UEBA.” It also explains that FortiSIEM agents can be installed on endpoints or servers to provide data collection functions that native syslog may not provide. For Windows systems specifically, the guide states that Windows servers do not natively send syslog messages and that a FortiSIEM Windows agent can be installed to perform that function. The FortiSIEM 7.4 User Guide also confirms that FortiInsight UEBA functionality runs as an integrated module within the FortiSIEM Windows Agent in newer releases. SSH and SNMP are access or monitoring protocols; they can support discovery or performance monitoring, but they are not the UEBA data-sending component. A FortiSIEM worker performs analysis and search functions inside the FortiSIEM architecture; it is not installed on endpoints to collect UEBA telemetry. Therefore, the FortiSIEM agent is the correct mechanism for sending UEBA-relevant endpoint data to FortiSIEM.

The correct answer is Local Auto. The uploaded answer was right, but its explanation was sloppy because it incorrectly described Local mode as the most time-consuming mode. In FortiSIEM machine learning, Local Auto mode selects the best algorithm by evaluating multiple candidate algorithms. The User Guide states that in Local Auto mode, “FortiSIEM picks the best algorithm” and that the Max Run Time parameter limits how long the job can run; longer runtime can produce better results. That is why Local Auto can take the most time. Forecasting and Regression are task types, not run modes.

The correct answer is B. Train. In FortiSIEM machine learning, the Train step is where the model is built from the prepared dataset and where model-fitting behavior can be adjusted. The FortiSIEM 7.4 User Guide explains that after preparing data, the analyst goes to Analytics > Machine Learning > Train, selects the machine learning task, chooses the algorithm, selects the prediction/target fields when required, and chooses the Train factor, which determines how much data is used for training versus testing. The guide states that the Train factor should be greater than 70%, meaning 70% of the data is used for training and 30% for testing. It also explains that model quality metrics show how accurately the algorithm predicts the field. For regression, lower MAE means a better fit, and R2 shows how well predictions approximate real data points. Most importantly, the guide states: “If you want to change the algorithm parameters and re-train, then click Tune & Train, change the parameters and click Save & Train.” This confirms that modifying how the model fits the training dataset is done in the Train step, not Prepare Data, Statistics, or Design.